Professional Documents
Culture Documents
table’
May 27, 2018
Version 1.11.4
Title Extension of `data.frame`
Depends R (>= 3.1.0)
Imports methods
Suggests bit64, curl, knitr, xts, nanotime, zoo, reshape2
Description Fast aggregation of large data (e.g. 100GB in RAM), fast or-
dered joins, fast add/modify/delete of columns by group us-
ing no copies at all, list columns, friendly and fast character-separated-value read/write. Of-
fers a natural and flexible syntax, for faster development.
License MPL-2.0 | file LICENSE
URL http://r-datatable.com
BugReports https://github.com/Rdatatable/data.table/issues
VignetteBuilder knitr
ByteCompile TRUE
NeedsCompilation yes
Author Matt Dowle [aut, cre],
Arun Srinivasan [aut],
Jan Gorecki [ctb],
Michael Chirico [ctb],
Pasha Stetsenko [ctb],
Tom Short [ctb],
Steve Lianoglou [ctb],
Eduard Antonyan [ctb],
Markus Bonsch [ctb],
Hugh Parsonage [ctb]
Maintainer Matt Dowle <mattjdowle@gmail.com>
Repository CRAN
Date/Publication 2018-05-27 16:34:37 UTC
1
2 R topics documented:
R topics documented:
data.table-package . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
:= . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
address . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
all.equal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
as.data.table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
as.data.table.xts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
as.matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
as.xts.data.table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
between . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
chmatch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
copy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
data.table-class . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
datatable.optimize . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
dcast.data.table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
duplicated . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
first . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
foverlaps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
frank . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
fread . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
fsort . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
fwrite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
groupingsets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
IDateTime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
J . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
last . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
like . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
melt.data.table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
merge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
na.omit.data.table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
patterns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
print.data.table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
rbindlist . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
rleid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
rowid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
setattr . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
setcolorder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
setDF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
setDT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
setDTthreads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
setkey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
setNumericRounding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
setops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
setorder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
shift . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
shouldPrint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
special-symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
data.table-package 3
split . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
subset.data.table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
test.data.table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
timetaken . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
transform.data.table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
transpose . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
truelength . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
tstrsplit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
update.dev.pkg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
Index 107
Description
data.table inherits from data.frame. It offers fast and memory efficient: file reader and writer,
aggregations, updates, equi, non-equi, rolling, range and interval joins, in a short and flexible syntax,
for faster development.
It is inspired by A[B] syntax in R where A is a matrix and B is a 2-column matrix. Since a
data.table is a data.frame, it is compatible with R functions and packages that accept only
data.frames.
Type vignette(package="data.table") to get started. The Introduction to data.table vignette
introduces data.table’s x[i, j, by] syntax and is a good place to start. If you have read the
vignettes and the help page below, please read the data.table support guide.
Please check the homepage for up to the minute live NEWS.
Tip: one of the quickest ways to learn the features is to type example(data.table) and study the
output at the prompt.
Usage
data.table(..., keep.rownames=FALSE, check.names=FALSE, key=NULL, stringsAsFactors=FALSE)
Arguments
... Just as ... in data.frame. Usual recycling rules are applied to vectors of
different lengths to create a list of equal length vectors.
keep.rownames If ... is a matrix or data.frame, TRUE will retain the rownames of that object
in a column named rn.
check.names Just as check.names in data.frame.
key Character vector of one or more column names which is passed to setkey. It
may be a single comma separated string such as key="x,y,z", or a vector of
names such as key=c("x","y","z").
stringsAsFactors
Logical (default is FALSE). Convert all character columns to factors?
x A data.table.
i Integer, logical or character vector, single column numeric matrix, expression
of column names, list, data.frame or data.table.
integer and logical vectors work the same way they do in [.data.frame
except logical NAs are treated as FALSE.
expression is evaluated within the frame of the data.table (i.e. it sees column
names as if they are variables) and can evaluate to any of the other types.
character, list and data.frame input to i is converted into a data.table
internally using as.data.table.
If i is a data.table, the columns in i to be matched against x can be specified
using one of these ways:
• on argument (see below). It allows for both equi- and the newly imple-
mented non-equi joins.
• If not, x must be keyed. Key can be set using setkey. If i is also keyed,
then first key column of i is matched against first key column of x, second
against second, etc..
If i is not keyed, then first column of i is matched against first key column
of x, second column of i against second key column of x, etc...
This is summarised in code as min(length(key(x)), if (haskey(i)) length(key(i)) else nco
Using on= is recommended (even during keyed joins) as it helps understand the
code better and also allows for non-equi joins.
When the binary operator == alone is used, an equi join is performed. In SQL
terms, x[i] then performs a right join by default. i prefixed with ! signals a
not-join or not-select.
Support for non-equi join was recently implemented, which allows for other
binary operators >=, >, <= and <.
See Keys and fast binary search based subset and Secondary indices and auto
indexing.
Advanced: When i is a single variable name, it is not considered an expression
of column names and is instead evaluated in calling scope.
j When with=TRUE (default), j is evaluated within the frame of the data.table;
i.e., it sees column names as if they are variables. This allows to not just select
columns in j, but also compute on them e.g., x[, a] and x[, sum(a)] returns
data.table-package 5
x$a and sum(x$a) as a vector respectively. x[, .(a, b)] and x[, .(sa=sum(a), sb=sum(b))]
returns a two column data.table each, the first simply selecting columns a, b
and the second computing their sums.
The expression ‘.()‘ is a shorthand alias to list(); they both mean the same.
(An exception is made for the use of .() within a call to bquote, where .() is
left unchanged.) As long as j returns a list, each element of the list becomes a
column in the resulting data.table. This is the default enhanced mode.
When with=FALSE, j can be a vector of column names or positions to select
(as in data.frame), or a logical vector with length ncol(x) defining columns
to select. Note: if a logical vector with length k < ncol(x) is passed, it will
be filled to length ncol(x) with FALSE, which is different from data.frame,
where the vector is recycled.
Advanced: j also allows the use of special read-only symbols: .SD, .N, .I,
.GRP, .BY.
Advanced: When i is a data.table, the columns of i can be referred to in j by
using the prefix i., e.g., X[Y, .(val, i.val)]. Here val refers to X’s column
and i.val Y’s.
Advanced: Columns of x can now be referred to using the prefix x. and is par-
ticularly useful during joining to refer to x’s join columns as they are otherwise
masked by i’s. For example, X[Y, .(x.a-i.a, b), on="a"].
See Introduction to data.table vignette and examples.
by Column names are seen as if they are variables (as in j when with=TRUE). The
data.table is then grouped by the by and j is evaluated within each group. The
order of the rows within each group is preserved, as is the order of the groups.
by accepts:
• A single unquoted column name: e.g., DT[, .(sa=sum(a)), by=x]
• a list() of expressions of column names: e.g., DT[, .(sa=sum(a)), by=.(x=x>0, y)]
• a single character string containing comma separated column names (where
spaces are significant since column names may contain spaces even at the
start or end): e.g., DT[, sum(a), by="x,y,z"]
• a character vector of column names: e.g., DT[, sum(a), by=c("x", "y")]
• or of the form startcol:endcol: e.g., DT[, sum(a), by=x:z]
Advanced: When i is a list (or data.frame or data.table), DT[i, j, by=.EACHI]
evaluates j for the groups in ‘DT‘ that each row in i joins to. That is, you can
join (in i) and aggregate (in j) simultaneously. We call this grouping by each i.
See this StackOverflow answer for a more detailed explanation until we roll out
vignettes.
Advanced: In the X[Y, j] form of grouping, the j expression sees variables in
X first, then Y. We call this join inherited scope. If the variable is not in X or Y
then the calling frame is searched, its calling frame, and so on in the usual way
up to and including the global environment.
keyby Same as by, but with an additional setkey() run on the by columns of the result,
for convenience. It is common practice to use ‘keyby=‘ routinely when you wish
the result to be sorted.
with By default with=TRUE and j is evaluated within the frame of x; column names
can be used as variables.
6 data.table-package
verbose TRUE turns on status and information messages to the console. Turn this on by
default using options(datatable.verbose=TRUE). The quantity and types of
verbosity may be expanded in future.
allow.cartesian
FALSE prevents joins that would result in more than nrow(x)+nrow(i) rows.
This is usually caused by duplicate values in i’s join columns, each of which
join to the same group in ‘x‘ over and over again: a misspecified join. Usually
this was not intended and the join needs to be changed. The word ’cartesian’
is used loosely in this context. The traditional cartesian join is (deliberately)
difficult to achieve in data.table: where every row in i joins to every row in
x (a nrow(x)*nrow(i) row result). ’cartesian’ is just meant in a ’large multi-
plicative’ sense.
drop Never used by data.table. Do not use. It needs to be here because data.table
inherits from data.frame. See datatable-faq.
on Indicate which columns in x should be joined with which columns in i along
with the type of binary operator to join with (see non-equi joins below on this).
When specified, this overrides the keys set on x and i. There are multiple ways
of specifying the on argument:
• As an unnamed character vector, e.g., X[Y, on=c("a", "b")], used when
columns a and b are common to both X and Y.
• Foreign key joins: As a named character vector when the join columns have
different names in X and Y. For example, X[Y, on=c(x1="y1", x2="y2")]
joins X and Y by matching columns x1 and x2 in X with columns y1 and y2
in Y, respectively.
From v1.9.8, you can also express foreign key joins using the binary oper-
ator ==, e.g. X[Y, on=c("x1==y1", "x2==y2")].
NB: shorthand like X[Y, on=c("a", V2="b")] is also possible if, e.g.,
column "a" is common between the two tables.
• For convenience during interactive scenarios, it is also possible to use .()
syntax as X[Y, on=.(a, b)].
• From v1.9.8, (non-equi) joins using binary operators >=, >, <=, < are
also possible, e.g., X[Y, on=c("x>=a", "y<=b")], or for interactive use
as X[Y, on=.(x>=a, y<=b)].
See examples as well as Secondary indices and auto indexing.
Details
The way to read this out loud is: "Take DT, subset rows by i, then compute j grouped by by. Here
are some basic usage examples expanding on this definition. See the vignette (and examples) for
working examples.
X[, a] # return col 'a' from X as vector. If not found, search in parent frame.
X[, .(a)] # same as above, but return as a data.table.
X[, sum(a)] # return sum(a) as a vector (with same scoping rules as above)
X[, .(sum(a)), by=c] # get sum(a) grouped by 'c'.
X[, sum(a), by=c] # same as above, .() can be omitted in by on single expression for conveni
X[, sum(a), by=c:f] # get sum(a) grouped by all columns in between 'c' and 'f' (both inclusiv
X[, sum(a), keyby=b] # get sum(a) grouped by 'b', and sort that result by the grouping column
X[, sum(a), by=b][order(b)] # same order as above, but by chaining compound expressions
X[c>1, sum(a), by=c] # get rows where c>1 is TRUE, and on those rows, get sum(a) grouped by 'c
X[Y, .(a, b), on="c"] # get rows where Y$c == X$c, and select columns 'X$a' and 'X$b' for those
X[Y, .(a, i.a), on="c"] # get rows where Y$c == X$c, and then select 'X$a' and 'Y$a' (=i.a)
X[Y, sum(a*i.a), on="c" by=.EACHI] # for *each* 'Y$c', get sum(a*i.a) on matching rows in 'X$c'
X[, plot(a, b), by=c] # j accepts any expression, generates plot for each group and returns no
# see ?assign to add/update/delete columns by reference using the same consistent interface
1. it never has or uses rownames. Rownames based indexing can be done by setting a key of one
or more columns or done ad-hoc using the on argument (now preferred).
2. it has enhanced functionality in [.data.table for fast joins of keyed tables, fast aggrega-
tion, fast last observation carried forward (LOCF) and fast add/modify/delete of columns by
reference with no copy at all.
See the see also section for the several other methods that are available for operating on data.tables
efficiently.
Note
If keep.rownames or check.names are supplied they must be written in full because R does not
allow partial argument names after ‘...‘. For example, data.table(DF, keep=TRUE) will create a
column called "keep" containing TRUE and this is correct behaviour; data.table(DF, keep.rownames=TRUE)
was intended.
POSIXlt is not supported as a column type because it uses 40 bytes to store a single datetime. They
are implicitly converted to POSIXct type with warning. You may also be interested in IDateTime
instead; it has methods to convert to and from POSIXlt.
data.table-package 9
References
https://github.com/Rdatatable/data.table/wiki (data.table homepage)
http://en.wikipedia.org/wiki/Binary_search
See Also
special-symbols, data.frame, [.data.frame, as.data.table, setkey, setorder, setDT, setDF,
J, SJ, CJ, merge.data.table, tables, test.data.table, IDateTime, unique.data.table, copy,
:=, alloc.col, truelength, rbindlist, setNumericRounding, datatable-optimize, fsetdiff,
funion, fintersect, fsetequal, anyDuplicated, uniqueN, rowid, rleid, na.omit, frank
Examples
## Not run:
example(data.table) # to run these examples at the prompt
## End(Not run)
is.data.frame(DT) # TRUE
tables()
# joins as subsets
X = data.table(x=c("c","b"), v=8:7, foo=c(4,2))
X
# setting keys
kDT = copy(DT) # (deep) copy DT to kDT to work with it.
setkey(kDT,x) # set a 1-column key. No quotes, for convenience.
setkeyv(kDT,"x") # same (v in setkeyv stands for vector)
v="x"
setkeyv(kDT,v) # same
# key(kDT)<-"x" # copies whole table, please use set* functions instead
haskey(kDT) # TRUE
key(kDT) # "x"
# all together
kDT[!"a", sum(v), by=.EACHI] # get sum(v) for each i != "a"
# multi-column key
setkey(kDT,x,y) # 2-column key
setkeyv(kDT,c("x","y")) # same
DT[, list(MySum=sum(v),
MyMin=min(v),
MyMax=max(v)),
by=.(x, y%%2)] # by 2 expressions
# using rleid, get max(y) and min of all cols in .SDcols for each consecutive run of 'v'
DT[, c(.(y=max(y)), lapply(.SD, min)), by=rleid(v), .SDcols=v:b]
## Not run:
if (interactive()) {
vignette("datatable-intro")
vignette("datatable-reference-semantics")
vignette("datatable-keys-fast-subset")
vignette("datatable-secondary-indices-and-auto-indexing")
vignette("datatable-reshape")
vignette("datatable-faq")
# get the latest devel version (compiled binary for Windows available -- no tools needed)
# https://github.com/Rdatatable/data.table/wiki/Installation
}
## End(Not run)
:= Assignment by reference
Description
Fast add, remove and update subsets of columns, by reference. := operator can be used in two ways:
LHS := RHS form, and Functional form. See Usage.
set is a low-overhead loop-able version of :=. It is particularly useful for repetitively updating rows
of certain columns by reference (using a for-loop). See Examples. It can not perform grouping
operations.
Usage
# 1. LHS := RHS form
# DT[i, LHS := RHS, by = ...]
# DT[i, c("LHS1", "LHS2") := list(RHS1, RHS2), by = ...]
# 2. Functional form
# DT[i, `:=`(LHS1 = RHS1,
# LHS2 = RHS2,
# ...), by = ...]
Arguments
LHS A character vector of column names (or numeric positions) or a variable that
evaluates as such. If the column doesn’t exist, it is added, by reference.
RHS A list of replacement values. It is recycled in the usual way to fill the number of
rows satisfying i, if any. To remove a column use NULL.
x A data.table. Or, set() accepts data.frame, too.
i Optional. Indicates the rows on which the values must be updated with. If not
provided, implies all rows. The := form is more powerful as it allows subsets
and joins based add/update columns by reference. See Details.
In set, only integer type is allowed in i indicating which rows value should
be assigned to. NULL represents all rows more efficiently than creating a vector
such as 1:nrow(x).
14 :=
Details
:= is defined for use in j only. It adds or updates or removes column(s) by reference. It makes no
copies of any part of memory at all. Read the Reference Semantics HTML vignette to follow with
examples. Some typical usages are:
DT[, col := val] # update (or add at the end if doesn't exist) a column call
DT[i, col := val] # same as above, but only for those rows specified in i and
DT[i, "col a" := val] # same. column is called "col a"
DT[i, (3:6) := val] # update existing columns 3:6 with value. Aside: parens ar
DT[i, colvector := val, with = FALSE] # OLD syntax. The contents of "colvector" in calling sc
DT[i, (colvector) := val] # same (NOW PREFERRED) shorthand syntax. The parens are e
DT[i, colC := mean(colB), by = colA] # update (or add) column called "colC" by reference by g
DT[,`:=`(new1 = sum(colB), new2 = sum(colC))] # Functional form
x := 1L
DT[i, col] := val
DT[i]$col := val
DT[, {col1 := 1L; col2 := 2L}] # Use the functional form, `:=`(), instead (see above).
For additional resources, check the FAQs vignette. Also have a look at StackOverflow’s data.table
tag.
:= in j can be combined with all types of i (such as binary search), and all types of by. This a one
reason why := has been implemented in j. See the Reference Semantics HTML vignette and also
FAQ 2.16 for analogies to SQL.
When LHS is a factor column and RHS is a character vector with items missing from the factor levels,
the new level(s) are automatically added (by reference, efficiently), unlike base methods.
Unlike <- for data.frame, the (potentially large) LHS is not coerced to match the type of the (often
small) RHS. Instead the RHS is coerced to match the type of the LHS, if necessary. Where this
involves double precision values being coerced to an integer column, a warning is given (whether
or not fractional data is truncated). The motivation for this is efficiency. It is best to get the column
types correct up front and stick to them. Changing a column type is possible but deliberately harder:
provide a whole column as the RHS. This RHS is then plonked into that column slot and we call this
plonk syntax, or replace column syntax if you prefer. By needing to construct a full length vector of
a new type, you as the user are more aware of what is happening, and it’s clearer to readers of your
code that you really do intend to change the column type.
data.tables are not copied-on-change by :=, setkey or any of the other set* functions. See
copy.
Value
DT is modified by reference and returned invisibly. If you require a copy, take a copy first (using
DT2 = copy(DT)).
:= 15
Advanced (internals):
It is easy to see how sub-assigning to existing columns is done internally. Removing columns by
reference is also straightforward by modifying the vector of column pointers only (using memmove
in C). However adding (new) columns is more tricky as to how the data.table can be grown by
reference: the list vector of column pointers is over-allocated, see truelength. By defining := in
j we believe update syntax is natural, and scales, but it also bypasses [<- dispatch and allows := to
update by reference with no copies of any part of memory at all.
Since [.data.table incurs overhead to check the existence and type of arguments (for example),
set() provides direct (but less flexible) assignment by reference with low overhead, appropriate
for use inside a for loop. See examples. := is more powerful and flexible than set() because := is
intended to be combined with i and by in single queries on large datasets.
Note:
DT[a > 4, b := c] is different from DT[a > 4][, b := c]. The first expression updates (or
adds) column b with the value c on those rows where a > 4 evaluates to TRUE. X is updated by
reference, therefore no assignment needed.
The second expression on the other hand updates a new data.table that’s returned by the subset
operation. Since the subsetted data.table is ephemeral (it is not assigned to a symbol), the result
would be lost; unless the result is assigned, for example, as follows: ans <- DT[a > 4][, b := c].
See Also
data.table, copy, alloc.col, truelength, set
Examples
DT = data.table(a = LETTERS[c(3L,1:3)], b = 4:7)
DT[, c := 8] # add a numeric column, 8 for all rows
DT[, d := 9L] # add an integer column, 9L for all rows
DT[, c := NULL] # remove column c
DT[2, d := -8L] # subassign by reference to d; 2nd row is -8L now
DT # DT changed by reference
DT[2, d := 10L][] # shorthand for update and print
DT[b > 4, b := d * 2L] # subassign to b with d*2L on those rows where b > 4 is TRUE
DT[b > 4][, b := d * 2L] # different from above. [, := ] is performed on the subset
# which is an new (ephemeral) data.table. Result needs to be
# assigned to a variable (using `<-`).
## Not run:
# Speed example ...
# However, normally, we call [.data.table *once* on *large* data, not many times on small data.
# The above is to demonstrate overhead, not to recommend looping in this way. But the option
# of set() is there if you need it.
## End(Not run)
Description
Returns the pointer address of its argument.
all.equal 17
Usage
address(x)
Arguments
x Anything.
Details
Sometimes useful in determining whether a value has been copied or not, programmatically.
Value
A character vector length 1.
References
http://stackoverflow.com/a/10913296/403310 (but implemented in C without using .Internal(inspect()))
Description
Convenient test of data equality between data.table objects. Performs some factor level stripping.
Usage
## S3 method for class 'data.table'
all.equal(target, current, trim.levels=TRUE, check.attributes=TRUE,
ignore.col.order=FALSE, ignore.row.order=FALSE, tolerance=sqrt(.Machine$double.eps),
...)
Arguments
target, current
data.tables to compare
trim.levels A logical indicating whether or not to remove all unused levels in columns that
are factors before running equality check. It effect only when check.attributes
is TRUE and ignore.row.order is FALSE.
check.attributes
A logical indicating whether or not to check attributes, will apply not only to
data.table but also attributes of the columns. It will skip c("row.names",".internal.selfref")
data.table attributes.
ignore.col.order
A logical indicating whether or not to ignore columns order in data.table.
18 all.equal
ignore.row.order
A logical indicating whether or not to ignore rows order in data.table. This
option requires datasets to use data types on which join can be made, so no
support for list, complex, raw, but still supports integer64.
tolerance A numeric value used when comparing numeric columns, by default sqrt(.Machine$double.eps).
Unless non-default value provided it will be forced to 0 if used together with
ignore.row.order and duplicate rows detected or factor columns present.
... Passed down to internal call of all.equal.
Details
For efficiency data.table method will exit on detected non-equality issues, unlike most all.equal
methods which process equality checks further. Besides that fact it also handles the most time
consuming case of ignore.row.order = TRUE very efficiently.
Value
Either TRUE or a vector of mode "character" describing the differences between target and
current.
See Also
all.equal
Examples
dt1 <- data.table(A = letters[1:10], X = 1:10, key = "A")
dt2 <- data.table(A = letters[5:14], Y = 1:10, key = "A")
isTRUE(all.equal(dt1, dt1))
is.character(all.equal(dt1, dt2))
# ignore.col.order
x <- copy(dt1)
y <- dt1[, .(X, A)]
all.equal(x, y)
all.equal(x, y, ignore.col.order = TRUE)
# ignore.row.order
x <- setkeyv(copy(dt1), NULL)
y <- dt1[sample(nrow(dt1))]
all.equal(x, y)
all.equal(x, y, ignore.row.order = TRUE)
# check.attributes
x = copy(dt1)
y = setkeyv(copy(dt1), NULL)
all.equal(x, y)
all.equal(x, y, check.attributes = FALSE)
# trim.levels
x <- data.table(A = factor(letters[1:10])[1:4]) # 10 levels
as.data.table 19
Description
Functions to check if an object is data.table, or coerce it if possible.
Usage
as.data.table(x, keep.rownames=FALSE, ...)
is.data.table(x)
Arguments
x An R object.
keep.rownames Default is FALSE. If TRUE, adds the input object’s names as a separate column
named "rn". keep.rownames = "id" names the column "id" instead.
sorted logical used in array method, default TRUE.
value.name character scalar used in array method, default "value".
na.rm logical used in array method, default TRUE will remove rows with NA values.
... Additional arguments to be passed to or from other methods.
Details
as.data.table is a generic function with many methods, and other packages can supply further
methods.
If a list is supplied, each element is converted to a column in the data.table with shorter ele-
ments recycled automatically. Similarly, each column of a matrix is converted separately.
character objects are not converted to factor types unlike as.data.frame.
If a data.frame is supplied, all classes preceding "data.frame" are stripped. Similarly, for
data.table as input, all classes preceding "data.table" are stripped. as.data.table methods
returns a copy of original data. To modify by reference see setDT and setDF.
keep.rownames argument can be used to preserve the (row)names attribute in the resulting data.table.
20 as.data.table.xts
See Also
data.table, setDT, setDF, copy, setkey, J, SJ, CJ, merge.data.table, :=, alloc.col, truelength,
rbindlist, setNumericRounding, datatable-optimize
Examples
nn = c(a=0.1, b=0.2, c=0.3, d=0.4)
as.data.table(nn)
as.data.table(nn, keep.rownames=TRUE)
as.data.table(nn, keep.rownames="rownames")
ll = list(a=1:2, b=3:4)
as.data.table(ll)
as.data.table(ll, keep.rownames=TRUE)
as.data.table(ll, keep.rownames="rownames")
DT = data.table(x=rep(c("x","y","z"),each=2), y=c(1:6))
as.data.table(DT)
ar = rnorm(27)
ar[sample(27, 15)] = NA
dim(ar) = c(3L,3L,3L)
as.data.table(ar)
Description
Efficient conversion xts to data.table.
Usage
## S3 method for class 'xts'
as.data.table(x, keep.rownames = TRUE, ...)
as.matrix 21
Arguments
See Also
as.xts.data.table
Examples
if (requireNamespace("xts", quietly = TRUE)) {
data(sample_matrix, package = "xts")
sample.xts <- xts::as.xts(sample_matrix) # xts might not be attached on search path
# print head of xts
print(head(sample.xts))
# print data.table
print(as.data.table(sample.xts))
}
Description
Converts a data.table into a matrix, optionally using one of the columns in the data.table as
the matrix rownames.
Usage
Arguments
x a data.table
rownames optional, a single column name or column index to use as the rownames in the
returned matrix. If TRUE the key of the data.table will be used if it is a single
column, otherwise the first column in the data.table will be used. Alternative
a vector of length nrow(x) to assign as the row names of the returned matrix.
... additional arguments to be passed to or from methods.
22 as.xts.data.table
Details
as.matrix is a generic function in base R. It dispatches to as.matrix.data.table if its x argu-
ment is a data.table.
The method for data.tables will return a character matrix if there are only atomic columns and any
non-(numeric/logical/complex) column, applying as.vector to factors and format to other non-
character columns. Otherwise, the usual coercion hierarchy (logical < integer < double < complex)
will be used, e.g., all-logical data frames will be coerced to a logical matrix, mixed logical-integer
will give an integer matrix, etc.
An additional argument rownames is provided for as.matrix.data.table to facilitate conversions
to matrices where the rownames are stored in a single column of x, e.g. the first column after using
dcast.data.table.
Value
A new matrix containing the contents of x.
See Also
data.table, as.matrix, data.matrix array
Examples
(dt1 <- data.table(A = letters[1:10], X = 1:10, Y = 11:20))
as.matrix(dt1) # character matrix
as.matrix(dt1, rownames = "A")
as.matrix(dt1, rownames = 1)
as.matrix(dt1, rownames = TRUE)
Description
Efficient conversion of data.table to xts, data.table must have POSIXct or Date type in first column.
Usage
as.xts.data.table(x, ...)
Arguments
x data.table to convert to xts, must have POSIXct or Date in the first column. All
others non-numeric columns will be omitted with warning.
... ignored, just for consistency with generic method.
between 23
See Also
as.data.table.xts
Examples
if (requireNamespace("xts", quietly = TRUE)) {
sample.dt <- data.table(date = as.Date((Sys.Date()-999):Sys.Date(),origin="1970-01-01"),
quantity = sample(10:50,1000,TRUE),
value = sample(100:1000,1000,TRUE))
# print data.table
print(sample.dt)
# print head of xts
print(head(as.xts.data.table(sample.dt))) # xts might not be attached on search path
}
Description
Intended for use in i in [.data.table.
between is equivalent to x >= lower & x <= upper when incbounds=TRUE, or x > lower & y < upper
when FALSE.
inrange checks whether each value in x is in between any of the intervals provided in lower,upper.
Usage
between(x, lower, upper, incbounds=TRUE)
x %between% y
Arguments
x Any orderable vector, i.e., those with relevant methods for `<=`, such as numeric,
character, Date, etc. in case of between and a numeric vector in case of
inrange.
lower Lower range bound.
upper Upper range bound.
y A length-2 vector or list, with y[[1]] interpreted as lower and y[[2]] as
upper.
incbounds TRUE means inclusive bounds, i.e., [lower,upper]. FALSE means exclusive bounds,
i.e., (lower,upper).
It is set to TRUE by default for infix notations.
24 chmatch
Details
From v1.9.8+, between is vectorised. lower and upper are recycled to length(x) if necessary.
non-equi joins were recently implemented in v1.9.8. It extends binary search based joins in
data.table to other binary operators including >=, <=, >, <. inrange makes use of this new
functionality and performs a range join.
Value
Logical vector as the same length as x with value TRUE for those that lie within the specified range.
Note
Current implementation does not make use of ordered keys for %between%.
See Also
Examples
X = data.table(a=1:5, b=6:10, c=c(5:1))
X[b %between% c(7,9)]
X[between(b, 7, 9)] # same as above
# NEW feature in v1.9.8, vectorised between
X[c %between% list(a,b)]
X[between(c, a, b)] # same as above
X[between(c, a, b, incbounds=FALSE)] # open interval
# inrange()
Y = data.table(a=c(8,3,10,7,-10), val=runif(5))
range = data.table(start = 1:5, end = 6:10)
Y[a %inrange% range]
Y[inrange(a, range$start, range$end)] # same as above
Y[inrange(a, range$start, range$end, incbounds=FALSE)] # open interval
Description
chmatch returns a vector of the positions of (first) matches of its first argument in its second. Both
arguments must be character vectors.
%chin% is like %in%, but for character vectors.
chmatch 25
Usage
chmatch(x, table, nomatch=NA_integer_)
x %chin% table
chorder(x)
chgroup(x)
Arguments
x character vector: the values to be matched, or the values to be ordered or grouped
table character vector: the values to be matched against.
nomatch the value to be returned in the case when no match is found. Note that it is
coerced to integer.
Details
Fast versions of match, %in% and order, optimised for character vectors. chgroup groups together
duplicated values but retains the group order (according the first appearance order of each group),
efficiently. They have been primarily developed for internal use by data.table, but have been exposed
since that seemed appropriate.
Strings are already cached internally by R (CHARSXP) and that is utilised by these functions. No
hash table is built or cached, so the first call is the same speed as subsequent calls. Essentially,
a counting sort (similar to base::sort.list(x,method="radix"), see setkey) is implemented
using the (almost) unused truelength of CHARSXP as the counter. Where R has used truelength
of CHARSXP (where a character value is shared by a variable name), the non zero truelengths
are stored first and reinstated afterwards. Each of the ch* functions implements a variation on
this theme. Remember that internally in R, length of a CHARSXP is the nchar of the string and
DATAPTR is the string itself.
Methods that do build and cache a hash table (such as the fastmatch package) are much faster
on subsequent calls (almost instant) but a little slower on the first. Therefore chmatch may be
particularly suitable for ephemeral vectors (such as local variables in functions) or tasks that are
only done once. Much depends on the length of x and table, how many unique strings each
contains, and whether the position of the first match is all that is required.
It may be possible to speed up fastmatch’s hash table build time by using the technique in data.table,
and we have suggested this to its author. If successful, fastmatch would then be fastest in all cases.
Value
As match and %in%. chorder and chgroup return an integer index vector.
Note
The name charmatch was taken by charmatch, hence chmatch.
See Also
match, %in%
26 copy
Examples
# Please type 'example(chmatch)' to run this and see timings on your machine
N = 1e5
# N is small here (1e5) to reduce runtime because every day CRAN runs all example sections in all
# manual pages (as well as the package's test suite) using latest R-devel to catch problems early.
# The comments here apply when N has been changed to 1e8. Timings were run on 2018-05-13 with
# R 3.5.0 and data.table 1.11.2.
u = as.character(as.hexmode(1:10000))
y = sample(u,N,replace=TRUE)
x = sample(u)
# With N=1e8 ...
system.time(a <- match(x,y)) # 4.6s
system.time(b <- chmatch(x,y)) # 1.8s
identical(a,b)
Description
In data.table parlance, all set* functions change their input by reference. That is, no copy is
made at all, other than temporary working memory, which is as large as one column. The only
other data.table operator that modifies input by reference is :=. Check out the See Also section
below for other set* function data.table provides.
copy() copies an entire object.
Usage
copy(x)
Arguments
x A data.table.
data.table-class 27
Details
data.table provides functions that operate on objects by reference and minimise full object copies
as much as possible. Still, it might be necessary in some situations to work on an object’s copy
which can be done using DT.copy <- copy(DT). It may also be sometimes useful before := (or
set) is used to subassign to a column by reference.
A copy() may be required when doing dt_names = names(DT). Due to R’s copy-on-modify,
dt_names still points to the same location in memory as names(DT). Therefore modifying DT by
reference now, say by adding a new column, dt_names will also get updated. To avoid this, one has
to explicitly copy: dt_names <- copy(names(DT)).
Value
Returns a copy of the object.
See Also
data.table, setkey, setDT, setDF, set :=, setorder, setattr, setnames
Examples
# Type 'example(copy)' to run these at prompt and browse output
DT = data.table(A=5:1,B=letters[5:1])
DT2 = copy(DT) # explicit copy() needed to copy a data.table
setkey(DT2,B) # now just changes DT2
identical(DT,DT2) # FALSE. DT and DT2 are now different tables
DT = data.table(A=5:1, B=letters[5:1])
nm1 = names(DT)
nm2 = copy(names(DT))
DT[, C := 1L]
identical(nm1, names(DT)) # TRUE, nm1 is also changed by reference
identical(nm2, names(DT)) # FALSE, nm2 is a copy, different from names(DT)
Description
A data.table can be used in S4 class definitions as either a parent class (inside a contains argu-
ment of setClass), or as an element of an S4 slot.
Author(s)
Steve Lianoglou
See Also
data.table
28 datatable.optimize
Examples
## Used in inheritence.
setClass('SuperDataTable', contains='data.table')
## Used in a slot
setClass('Something', representation(x='character', dt='data.table'))
x <- new("Something", x='check', dt=data.table(a=1:10, b=11:20))
Description
data.table internally optimises certain expressions in order to improve performance. This section
briefly summarises those optimisations.
Note that there’s no additional input needed from the user to take advantage of these optimisations.
They happen automatically.
Run the code under the example section to get a feel for the performance benefits from these opti-
misations.
Details
data.table reads the global option datatable.optimize to figure out what level of optimisation
is required. The default value Inf activates all available optimisations.
At optimisation level >= 1, i.e., getOption("datatable.optimize") >= 1, these are the optimi-
sations:
• The base function order is internally replaced with data.table’s fast ordering. That is,
DT[order(...)] gets internally optimised to DT[forder(...)].
• The expression DT[, lapply(.SD, fun), by=.] gets optimised to DT[, list(fun(a), fun(b), ...), by=.]
where a,b, ... are columns in .SD. This improves performance tremendously.
• Similarly, the expression DT[, c(.N, lapply(.SD, fun)), by=.] gets optimised to
DT[, list(.N, fun(a), fun(b), ...)]. .N is just for example here.
• base::mean function is internally optimised to use data.table’s fastmean function. mean()
from base is an S3 generic and gets slow with many groups.
• Expressions in j which contain only the functions min, max, mean, median, var, sd, sum, prod, first, last, h
(for example, DT[, list(mean(x), median(x), min(y), max(y)), by=z]), they are very
effectively optimised using what we call GForce. These functions are automatically replaced
with a corresponding GForce version with pattern g*, e.g., prod becomes gprod.
Normally, once the rows belonging to each group are identified, the values corresponding to
the group are gathered and the j-expression is evaluated. This can be improved by computing
datatable.optimize 29
the result directly without having to gather the values or evaluating the expression for each
group (which can get costly with large number of groups) by implementing it specifically for
a particular function. As a result, it is extremely fast.
• In addition to all the functions above, ‘.N‘ is also optimised to use GForce, when used sep-
arately or when combined with the functions mentioned above. Note further that GForce-
optimized functions must be used separately, i.e., code like DT[ , max(x) - min(x), by=z]
will not currently be optimized to use gmax, gmin.
• Expressions of the form DT[i, j, by] are also optimised when i is a subset operation and j
is any/all of the functions discussed above.
• Supported operators: ==, %in%. Non-equi operators(>, <, etc.) are not supported yet because
non-equi joins are slower than vector based subsets.
• Queries on multiple columns are supported, if the connector is ’&’, e.g. DT[x == 2 & y == 3]
is supported, but DT[x == 2 | y == 3] is not.
• Optimization will currently be turned off when doing subset when cross product of elements
provided to filter on exceeds > 1e4. This most likely happens if multiple %in%, or %chin%
queries are combined, e.g. DT[x %in% 1:100 & y %in% 1:200] will not be optimized since
100 * 200 = 2e4 > 1e4.
• Queries with multiple criteria on one column are not supported, e.g. DT[x == 2 & x %in% c(2,5)]
is not supported.
• Queries with non-missing j are supported, e.g. DT[x == 3 & y == 5, .(new = x-y)]
or DT[x == 3 & y == 5, new := x-y] are supported. Also extends to queries using
with = FALSE.
• "notjoin" queries, i.e. queries that start with !, are only supported if there are no & connections,
e.g. DT[!x==3] is supported, but DT[!x==3 & y == 4] is not.
If in doubt, whether your query benefits from optimization, call it with the verbose = TRUE
argument. You should see "Optimized subsetting...".
Auto indexing: In case a query is optimized, but no appropriate key or index is found, data.table
automatically creates an index on the first run. Any successive subsets on the same column then
reuse this index to binary search (instead of vector scan) and is therefore fast. Auto indexing can
be switched off with the global option options(datatable.auto.index = FALSE). To switch off
using existing indices set global option options(datatable.use.index = FALSE).
See Also
setNumericRounding, getNumericRounding
Examples
## Not run:
# Generate a big data.table with a relatively many columns
30 datatable.optimize
set.seed(1L)
DT = lapply(1:20, function(x) sample(c(-100:100), 5e6L, TRUE))
setDT(DT)[, id := sample(1e5, 5e6, TRUE)]
print(object.size(DT), units="Mb") # 400MB, not huge, but will do
# 'order' optimisation
options(datatable.optimize = 1L) # optimisation 'on'
system.time(ans1 <- DT[order(id)])
options(datatable.optimize = 0L) # optimisation 'off'
system.time(ans2 <- DT[order(id)])
identical(ans1, ans2)
# optimisation of 'mean'
options(datatable.optimize = 1L) # optimisation 'on'
system.time(ans1 <- DT[, lapply(.SD, mean), by=id])
system.time(ans2 <- DT[, lapply(.SD, base::mean), by=id])
identical(ans1, ans2)
# GForce
options(datatable.optimize = 2L) # optimisation 'on'
system.time(ans1 <- DT[, lapply(.SD, median), by=id])
system.time(ans2 <- DT[, lapply(.SD, function(x) as.numeric(stats::median(x))), by=id])
identical(ans1, ans2)
# optimized subsets
options(datatable.optimize = 2L)
system.time(ans1 <- DT[id == 100L]) # vector scan
system.time(ans2 <- DT[id == 100L]) # vector scan
system.time(DT[id %in% 100:500]) # vector scan
options(datatable.optimize = 3L)
system.time(ans1 <- DT[id == 100L]) # index + binary search subset
system.time(ans2 <- DT[id == 100L]) # only binary search subset
system.time(DT[id %in% 100:500]) # only binary search subset again
## End(Not run)
dcast.data.table 31
Description
dcast.data.table is a much faster version of reshape2::dcast, but for data.tables. More
importantly, it’s capable of handling very large data quite efficiently in terms of memory usage in
comparison to reshape2::dcast.
From 1.9.6, dcast is implemented as an S3 generic in data.table. To melt or cast data.tables,
it is not necessary to load reshape2 any more. If you have load reshape2, do so before loading
data.table to prevent unwanted masking.
NEW: dcast.data.table can now cast multiple value.var columns and also accepts multiple
functions to fun.aggregate. See Examples for more.
Usage
## S3 method for class 'data.table'
dcast(data, formula, fun.aggregate = NULL, sep = "_",
..., margins = NULL, subset = NULL, fill = NULL,
drop = TRUE, value.var = guess(data),
verbose = getOption("datatable.verbose"))
Arguments
data A data.table.
formula A formula of the form LHS ~ RHS to cast, see Details.
fun.aggregate Should the data be aggregated before casting? If the formula doesn’t identify
a single observation for each cell, then aggregation defaults to length with a
message.
NEW: it is possible to provide a list of functions to fun.aggregate. See Ex-
amples.
sep Character vector of length 1, indicating the separating character in variable
names generated during casting. Default is _ for backwards compatibility.
... Any other arguments that may be passed to the aggregating function.
margins Not implemented yet. Should take variable names to compute margins on. A
value of TRUE would compute all margins.
subset Specified if casting should be done on a subset of the data. Ex: subset = .(col1 <= 5)
or subset = .(variable != "January").
fill Value with which to fill missing cells. If fun.aggregate is present, takes the
value by applying the function on a 0-length vector.
drop FALSE will cast by including all missing combinations.
NEW: Following #1512, c(FALSE, TRUE) will only include all missing com-
binations of formula LHS. And c(TRUE, FALSE) will only include all missing
combinations of formula RHS. See Examples.
32 dcast.data.table
value.var Name of the column whose values will be filled to cast. Function ‘guess()‘ tries
to, well, guess this column automatically, if none is provided.
NEW: it is now possible to cast multiple value.var columns simultaneously.
See Examples.
verbose Not used yet. May be dropped in the future or used to provide informative
messages through the console.
Details
The cast formula takes the form LHS ~ RHS, ex: var1 + var2 ~ var3. The order of entries in
the formula is essential. There are two special variables: . and .... . represents no variable; ...
represents all variables not otherwise mentioned in formula; see Examples.
dcast also allows value.var columns of type list.
When variable combinations in formula doesn’t identify a unique value in a cell, fun.aggregate
will have to be specified, which defaults to length if unspecified. The aggregating function should
take a vector as input and return a single value (or a list of length one) as output. In cases where
value.var is a list, the function should be able to handle a list input and provide a single value or
list of length one as output.
If the formula’s LHS contains the same column more than once, ex: dcast(DT, x+x~ y), then
the answer will have duplicate names. In those cases, the duplicate names are renamed using
make.unique so that key can be set without issues.
Names for columns that are being cast are generated in the same order (separated by an underscore,
_) from the (unique) values in each column mentioned in the formula RHS.
From v1.9.4, dcast tries to preserve attributes wherever possible.
NEW: From v1.9.6, it is possible to cast multiple value.var columns and also cast by provid-
ing multiple fun.aggregate functions. Multiple fun.aggregate functions should be provided
as a list, for e.g., list(mean, sum, function(x) paste(x, collapse=""). value.var can
be either a character vector or list of length=1, or a list of length equal to length(fun.aggregate).
When value.var is a character vector or a list of length 1, each function mentioned under fun.aggregate
is applied to every column specified under value.var column. When value.var is a list of length
equal to length(fun.aggregate) each element of fun.aggregate is applied to each element of
value.var column.
Value
A keyed data.table that has been cast. The key columns are equal to the variables in the formula
LHS in the same order.
See Also
melt.data.table, rowid, https://cran.r-project.org/package=reshape
Examples
require(data.table)
names(ChickWeight) <- tolower(names(ChickWeight))
DT <- melt(as.data.table(ChickWeight), id=2:4) # calls melt.data.table
dcast.data.table 33
# using subset
dcast(DT, chick ~ time, fun=mean, subset=.(time < 10 & chick < 20))
## Not run:
# benchmark against reshape2's dcast, minimum of 3 runs
set.seed(45)
DT <- data.table(aa=sample(1e4, 1e6, TRUE),
bb=sample(1e3, 1e6, TRUE),
cc = sample(letters, 1e6, TRUE), dd=runif(1e6))
system.time(dcast(DT, aa ~ cc, fun=sum)) # 0.12 seconds
system.time(dcast(DT, bb ~ cc, fun=mean)) # 0.04 seconds
# reshape2::dcast takes 31 seconds
system.time(dcast(DT, aa + bb ~ cc, fun=sum)) # 1.2 seconds
## End(Not run)
Description
duplicated returns a logical vector indicating which rows of a data.table are duplicates of a row
with smaller subscripts.
unique returns a data.table with duplicated rows removed, by columns specified in by argument.
When no by then duplicated rows by all columns are removed.
anyDuplicated returns the index i of the first duplicated entry if there is one, and 0 otherwise.
uniqueN is equivalent to length(unique(x)) when x is an atomic vector, and nrow(unique(x))
when x is a data.frame or data.table. The number of unique rows are computed directly without
materialising the intermediate unique data.table and is therefore faster and memory efficient.
Usage
## S3 method for class 'data.table'
duplicated(x, incomparables=FALSE, fromLast=FALSE, by=seq_along(x), ...)
Arguments
x A data.table. uniqueN accepts atomic vectors and data.frames as well.
... Not used at this time.
incomparables Not used. Here for S3 method consistency.
fromLast logical indicating if duplication should be considered from the reverse side, i.e.,
the last (or rightmost) of identical elements would correspond to duplicated = FALSE.
by character or integer vector indicating which combinations of columns from
x to use for uniqueness checks. By default all columns are being used. That was
changed recently for consistency to data.frame methods. In version < 1.9.8
default was key(x).
na.rm Logical (default is FALSE). Should missing values (including NaN) be removed?
duplicated 35
Details
Because data.tables are usually sorted by key, tests for duplication are especially quick when only
the keyed columns are considered. Unlike unique.data.frame, paste is not used to ensure equal-
ity of floating point data. It is instead accomplished directly and is therefore quite fast. data.table
provides setNumericRounding to handle cases where limitations in floating point representation is
undesirable.
v1.9.4 introduces anyDuplicated method for data.tables and is similar to base in functionality. It
also implements the logical argument fromLast for all three functions, with default value FALSE.
Value
duplicated returns a logical vector of length nrow(x) indicating which rows are duplicates.
unique returns a data table with duplicated rows removed.
anyDuplicated returns a integer value with the index of first duplicate. If none exists, 0L is re-
turned.
uniqueN returns the number of unique elements in the vector, data.frame or data.table.
See Also
setNumericRounding, data.table, duplicated, unique, all.equal, fsetdiff, funion, fintersect,
fsetequal
Examples
DT <- data.table(A = rep(1:3, each=4), B = rep(1:4, each=3),
C = rep(1:2, 6), key = "A,B")
duplicated(DT)
unique(DT)
duplicated(DT, by="B")
unique(DT, by="B")
# fromLast=TRUE
36 first
# anyDuplicated
anyDuplicated(DT, by=c("A", "B")) # 3L
any(duplicated(DT, by=c("A", "B"))) # TRUE
# uniqueN's na.rm=TRUE
x = sample(c(NA, NaN, runif(3)), 10, TRUE)
uniqueN(x, na.rm = FALSE) # 5, default
uniqueN(x, na.rm=TRUE) # 3
Description
Returns the first item of a vector or list, or the first row of a data.frame or data.table.
Usage
first(x, ...)
Arguments
x A vector, list, data.frame or data.table. Otherwise the S3 method of xts::first
is deployed.
... Not applicable for data.table::first. Any arguments here are passed through
to xts::first.
Value
If no other arguments are supplied it depends on the type of x. The first item of a vector or list. The
first row of a data.frame or data.table. Otherwise, whatever xts::first returns (if package
xts has been loaded, otherwise a helpful error).
If any argument is supplied in addition to x (such as n or keep in xts::first), regardless of x’s
type, then xts::first is called if xts has been loaded, otherwise a helpful error.
foverlaps 37
See Also
NROW, head, tail, last
Examples
first(1:5) # [1] 1
x = data.table(x=1:5, y=6:10)
first(x) # same as x[1]
Description
A fast binary-search based overlap join of two data.tables. This is very much inspired by
findOverlaps function from the Bioconductor package IRanges (see link below under See Also).
Usually, x is a very large data.table with small interval ranges, and y is much smaller keyed data.table
with relatively larger interval spans. For a usage in genomics, see the examples section.
NOTE: This is still under development, meaning it’s stable, but some features are yet to be imple-
mented. Also, some arguments and/or the function name itself could be changed.
Usage
foverlaps(x, y, by.x = if (!is.null(key(x))) key(x) else key(y),
by.y = key(y), maxgap = 0L, minoverlap = 1L,
type = c("any", "within", "start", "end", "equal"),
mult = c("all", "first", "last"),
nomatch = getOption("datatable.nomatch"),
which = FALSE, verbose = getOption("datatable.verbose"))
Arguments
x, y data.tables. y needs to be keyed, but not necessarily x. See examples.
by.x, by.y A vector of column names (or numbers) to compute the overlap joins. The last
two columns in both by.x and by.y should each correspond to the start and
end interval columns in x and y respectively. And the start column should
always be <= end column. If x is keyed, by.x is equal to key(x), else key(y).
by.y defaults to key(y).
maxgap It should be a non-negative integer value, >= 0. Default is 0 (no gap). For inter-
vals [a,b] and [c,d], where a<=b and c<=d, when c > b or d < a, the two
intervals don’t overlap. If the gap between these two intervals is <= maxgap,
these two intervals are considered as overlapping. Note: This is not yet imple-
mented.
38 foverlaps
minoverlap It should be a positive integer value, > 0. Default is 1. For intervals [a,b]
and [c,d], where a<=b and c<=d, when c<=b and d>=a, the two intervals over-
lap. If the length of overlap between these two intervals is >= minoverlap,
then these two intervals are considered to be overlapping. Note: This is not yet
implemented.
type Default value is any. Allowed values are any, within, start, end and equal.
Note: equal is not yet implemented. But this is just a normal join of the type
y[x, ...], unless you require also using maxgap and minoverlap arguments.
The types shown here are identical in functionality to the function findOverlaps
in the bioconductor package IRanges. Let [a,b] and [c,d] be intervals in x
and y with a<=b and c<=d. For type="start", the intervals overlap iff a == c.
For type="end", the intervals overlap iff b == d. For type="within", the
intervals overlap iff a>=c and b<=d. For type="equal", the intervals overlap
iff a==c and b==d. For type="any", as long as c<=b and d>=a, they over-
lap. In addition to these requirements, they also have to satisfy the minoverlap
argument as explained above.
NB: maxgap argument, when > 0, is to be interpreted according to the type of
the overlap. This will be updated once maxgap is implemented.
mult When multiple rows in y match to the row in x, mult=. controls which values
are returned - "all" (default), "first" or "last".
nomatch Same as nomatch in match. When a row (with interval say, [a,b]) in x has
no match in y, nomatch=NA (default) means NA is returned for y’s non-by.y
columns for that row of x. nomatch=0 means no rows will be returned for that
row of x. The default value (used when nomatch is not supplied) can be changed
from NA to 0 using options(datatable.nomatch=0).
which When TRUE, if mult="all" returns a two column data.table with the first
column corresponding to x’s row number and the second corresponding to y’s.
when nomatch=NA, no matches return NA for y, and if nomatch=0, those rows
where no match is found will be skipped; if mult="first" or "last", a
vector of length equal to the number of rows in x is returned, with no-match
entries filled with NA or 0 corresponding to the nomatch argument. Default is
FALSE, which returns a join with the rows in y.
verbose TRUE turns on status and information messages to the console. Turn this on by
default using options(datatable.verbose=TRUE). The quantity and types of
verbosity may be expanded in future.
Details
Very briefly, foverlaps() collapses the two-column interval in y to one-column of unique values
to generate a lookup table, and then performs the join depending on the type of overlap, using
the already available binary search feature of data.table. The time (and space) required to
generate the lookup is therefore proportional to the number of unique values present in the interval
columns of y when combined together.
Overlap joins takes advantage of the fact that y is sorted to speed-up finding overlaps. Therefore y
has to be keyed (see ?setkey) prior to running foverlaps(). A key on x is not necessary, although
it might speed things further. The columns in by.x argument should correspond to the columns
specified in by.y. The last two columns should be the interval columns in both by.x and by.y.
foverlaps 39
The first interval column in by.x should always be <= the second interval column in by.x, and
likewise for by.y. The storage.mode of the interval columns must be either double or integer.
It therefore works with bit64::integer64 type as well.
The lookup generation step could be quite time consuming if the number of unique values in y
are too large (ex: in the order of tens of millions). There might be improvements possible by
constructing lookup using RLE, which is a pending feature request. However most scenarios will
not have too many unique values for y.
Value
A new data.table by joining over the interval columns (along with other additional identifier
columns) specified in by.x and by.y.
NB: When which=TRUE: a) mult="first" or "last" returns a vector of matching row numbers
in y, and b) when mult="all" returns a data.table with two columns with the first containing row
numbers of x and the second column with corresponding row numbers of y.
nomatch=NA or 0 also influences whether non-matching rows are returned or not, as explained
above.
See Also
data.table, http://www.bioconductor.org/packages/release/bioc/html/IRanges.html, setNumericRounding
Examples
require(data.table)
## simple example:
x = data.table(start=c(5,31,22,16), end=c(8,50,25,18), val2 = 7:10)
y = data.table(start=c(10, 20, 30), end=c(15, 35, 45), val1 = 1:3)
setkey(y, start, end)
foverlaps(x, y, type="any", which=TRUE) ## return overlap indices
foverlaps(x, y, type="any") ## return overlap join
foverlaps(x, y, type="any", mult="first") ## returns only first match
foverlaps(x, y, type="within") ## matches iff 'x' is within 'y'
Description
Similar to base::rank but much faster. And it accepts vectors, lists, data.frames or data.tables
as input. In addition to the ties.method possibilities provided by base::rank, it also provides
ties.method="dense".
bit64::integer64 type is also supported.
Usage
Arguments
x A vector, or list with all it’s elements identical in length or data.frame or data.table.
... Only for lists, data.frames and data.tables. The columns to calculate ranks based
on. Do not quote column names. If ... is missing, all columns are considered by
default. To sort by a column in descending order prefix a "-", e.g., frank(x, a, -b,
c). The -b works when b is of type character as well.
cols A character vector of column names (or numbers) of x, to which obtain ranks
for.
order An integer vector with only possible values of 1 and -1, corresponding to as-
cending and descending order. The length of order must be either 1 or equal to
that of cols. If length(order) == 1, it’s recycled to length(cols).
na.last Control treatment of NAs. If TRUE, missing values in the data are put last; if
FALSE, they are put first; if NA, they are removed; if "keep" they are kept with
rank NA.
ties.method A character string specifying how ties are treated, see Details.
frank 41
Details
To be consistent with other data.table operations, NAs are considered identical to other NAs (and
NaNs to other NaNs), unlike base::rank. Therefore, for na.last=TRUE and na.last=FALSE, NAs
(and NaNs) are given identical ranks, unlike rank.
frank is not limited to vectors. It accepts data.tables (and lists and data.frames) as well. It accepts
unquoted column names (with names preceded with a - sign for descending order, even on character
vectors), for e.g., frank(DT, a, -b, c, ties.method="first") where a,b,c are columns in
DT. The equivalent in frankv is the order argument.
In addition to the ties.method values possible using base’s rank, it also provides another addi-
tional argument "dense" which returns the ranks without any gaps in the ranking. See examples.
Value
A numeric vector of length equal to NROW(x) (unless na.last = NA, when missing values are
removed). The vector is of integer type unless ties.method = "average" when it is of double
type (irrespective of ties).
See Also
data.table, setkey, setorder
Examples
# on vectors
x = c(4, 1, 4, NA, 1, NA, 4)
# NAs are considered identical (unlike base R)
# default is average
frankv(x) # na.last=TRUE
frankv(x, na.last=FALSE)
# ties.method = min
frankv(x, ties.method="min")
# ties.method = dense
frankv(x, ties.method="dense")
# on data.table
DT = data.table(x, y=c(1, 1, 1, 0, NA, 0, 2))
frankv(DT, cols="x") # same as frankv(x) from before
frankv(DT, cols="x", na.last="keep")
frankv(DT, cols="x", ties.method="dense", na.last=NA)
frank(DT, x, ties.method="dense", na.last=NA) # equivalent of above using frank
# on both columns
frankv(DT, ties.method="first", na.last="keep")
frank(DT, ties.method="first", na.last="keep") # equivalent of above using frank
# order argument
frank(DT, x, -y, ties.method="first")
# equivalent of above using frankv
frankv(DT, order=c(1L, -1L), ties.method="first")
42 fread
Description
Similar to read.table but faster and more convenient. All controls such as sep, colClasses and
nrows are automatically detected. bit64::integer64 types are also detected and read directly
without needing to read as character before converting.
Dates are read as character currently. They can be converted afterwards using the excellent fasttime
package or standard base functions.
‘fread‘ is for regular delimited files; i.e., where every row has the same number of columns. In
future, secondary separator (sep2) may be specified within each column. Such columns will be
read as type list where each cell is itself a vector.
Usage
Arguments
input Either the file name to read (containing no \n character), a shell command that
pre-processes the file (e.g. fread("grep blah filename")) or the input itself
as a string (containing at least one \n), see examples. In both cases, a length 1
character string. A filename input is passed through path.expand for conve-
nience. input can also be a URL starting with http:// or file://; see Details.
sep The separator between columns. Defaults to the character in the set [,\t |;:]
that separates the sample of rows into the most number of lines with the same
number of fields. Use NULL or "" to specify no separator; i.e. each line a single
character column like base::readLines does.
fread 43
sep2 The separator within columns. A list column will be returned where each
cell is a vector of values. This is much faster using less working memory than
strsplit afterwards or similar techniques. For each column sep2 can be dif-
ferent and is the first character in the same set above [,\t |;], other than sep,
that exists inside each field outside quoted regions in the sample. NB: sep2 is
not yet implemented.
nrows The maximum number of rows to read. Unlike read.table, you do not need to
set this to an estimate of the number of rows in the file for better speed because
that is already automatically determined by fread almost instantly using the
large sample of lines. ‘nrows=0‘ returns the column names and typed empty
columns determined by the large sample; useful for a dry run of a large file or to
quickly check format consistency of a set of files before starting to read any of
them.
header Does the first data line contain column names? Defaults according to whether
every non-empty field on the first data line is type character. If so, or TRUE is
supplied, any empty column names are given a default name.
na.strings A character vector of strings which are to be interpreted as NA values. By default,
",," for columns of all types, including type ‘character‘ is read as NA for consis-
tency. ,"", is unambiguous and read as an empty string. To read ,NA, as NA, set
na.strings="NA". To read ,, as blank string "", set na.strings=NULL. When
they occur in the file, the strings in na.strings should not appear quoted since
that is how the string literal ,"NA", is distinguished from ,NA,, for example,
when na.strings="NA".
file File path, useful when we want to ensure that no shell commands will be exe-
cuted. File path can also be provided to input argument.
stringsAsFactors
Convert all character columns to factors?
verbose Be chatty and report timings?
skip If 0 (default) start on the first line and from there finds the first row with a consis-
tent number of columns. This automatically avoids irregular header information
before the column names row. skip>0 means ignore the first skip rows man-
ually. skip="string" searches for "string" in the file (e.g. a substring of
the column names row) and starts on that line (inspired by read.xls in package
gdata).
select Vector of column names or numbers to keep, drop the rest.
drop Vector of column names or numbers to drop, keep the rest.
colClasses A character vector of classes (named or unnamed), as read.csv. Or a named list
of vectors of column names or numbers, see examples. colClasses in fread is
intended for rare overrides, not for routine use. fread will only promote a column
to a higher type if colClasses requests it. It won’t downgrade a column to a
lower type since NAs would result. You have to coerce such columns afterwards
yourself, if you really require data loss.
integer64 "integer64" (default) reads columns detected as containing integers larger than
2^31 as type bit64::integer64. Alternatively, "double"|"numeric" reads as
base::read.csv does; i.e., possibly with loss of precision and if so silently. Or,
"character".
44 fread
dec The decimal separator as in base::read.csv. If not "." (default) then usually
",". See details.
col.names A vector of optional names for the variables (columns). The default is to use
the header column if present or detected, or if not "V" followed by the column
number. This is applied after check.names and before key and index.
check.names default is FALSE. If TRUE then the names of the variables in the data.table are
checked to ensure that they are syntactically valid variable names. If necessary
they are adjusted (by make.names) so that they are, and also to ensure that there
are no duplicates.
encoding default is "unknown". Other possible options are "UTF-8" and "Latin-1".
Note: it is not used to re-encode the input, rather enables handling of encoded
strings in their native encoding.
quote By default ("\""), if a field starts with a double quote, fread handles embedded
quotes robustly as explained under Details. If it fails, then another attempt is
made to read the field as is, i.e., as if quotes are disabled. By setting quote="",
the field is always read as if quotes are disabled. It is not expected to ever need
to pass anything other than \"\" to quote; i.e., to turn it off.
strip.white default is TRUE. Strips leading and trailing whitespaces of unquoted fields. If
FALSE, only header trailing spaces are removed.
fill logical (default is FALSE). If TRUE then in case the rows have unequal length,
blank fields are implicitly filled.
blank.lines.skip
logical, default is FALSE. If TRUE blank lines in the input are ignored.
key Character vector of one or more column names which is passed to setkey. It
may be a single comma separated string such as key="x,y,z", or a vector of
names such as key=c("x","y","z"). Only valid when argument data.table=TRUE.
Where applicable, this should refer to column names given in col.names.
index Character vector or list of character vectors of one or more column names which
is passed to setindexv. As with key, comma-separated notation like index="x,y,z"
is accepted for convenience. Only valid when argument data.table=TRUE.
Where applicable, this should refer to column names given in col.names.
showProgress TRUE displays progress on the console if the ETA is greater than 3 seconds. It is
produced in fread’s C code where the very nice (but R level) txtProgressBar and
tkProgressBar are not easily available.
data.table TRUE returns a data.table. FALSE returns a data.frame.
nThread The number of threads to use. Experiment to see what works best for your data
on your hardware.
logical01 If TRUE a column containing only 0s and 1s will be read as logical, otherwise
as integer.
autostart Deprecated and ignored with warning. Please use skip instead.
Details
A sample of 10,000 rows is used for a very good estimate of column types. 100 contiguous rows
are read from 100 equally spaced points throughout the file including the beginning, middle and the
fread 45
very end. This results in a better guess when a column changes type later in the file (e.g. blank
at the beginning/only populated near the end, or 001 at the start but 0A0 later on). This very good
type guess enables a single allocation of the correct type up front once for speed, memory efficiency
and convenience of avoiding the need to set colClasses after an error. Even though the sample is
large and jumping over the file, it is almost instant regardless of the size of the file because a lazy
on-demand memory map is used. If a jump lands inside a quoted field containing newlines, each
newline is tested until 5 lines are found following it with the expected number of fields. The lowest
type for each column is chosen from the ordered list: logical, integer, integer64, double,
character. Rarely, the file may contain data of a higher type in rows outside the sample (referred
to as an out-of-sample type exception). In this event fread will automatically reread just those
columns from the beginning so that you don’t have the inconvenience of having to set colClasses
yourself; particularly helpful if you have a lot of columns. Such columns must be read from the
beginning to correctly distinguish "00" from "000" when those have both been interpreted as integer
0 due to the sample but 00A occurs out of sample. Set verbose=TRUE to see a detailed report of the
logic deployed to read your file.
There is no line length limit, not even a very large one. Since we are encouraging list columns (i.e.
sep2) this has the potential to encourage longer line lengths. So the approach of scanning each line
into a buffer first and then rescanning that buffer is not used. There are no buffers used in fread’s
C code at all. The field width limit is limited by R itself: the maximum width of a character string
(currently 2^31-1 bytes, 2GB).
The filename extension (such as .csv) is irrelevant for "auto" sep and sep2. Separator detection is
entirely driven by the file contents. This can be useful when loading a set of different files which
may not be named consistently, or may not have the extension .csv despite being csv. Some datasets
have been collected over many years, one file per day for example. Sometimes the file name format
has changed at some point in the past or even the format of the file itself. So the idea is that you can
loop fread through a set of files and as long as each file is regular and delimited, fread can read
them all. Whether they all stack is another matter but at least each one is read quickly without you
needing to vary colClasses in read.table or read.csv.
If an empty line is encountered then reading stops there with warning if any text exists after the
empty line such as a footer. The first line of any text discarded is included in the warning message.
Unless, it is single-column input. In that case blank lines are significant (even at the very end) and
represent NA in the single column. So that fread(fwrite(DT))==DT. This default behaviour can
be controlled using blank.lines.skip=TRUE|FALSE.
Line endings: All known line endings are detected automatically: \n (*NIX including Mac), \r\n
(Windows CRLF), \r (old Mac) and \n\r (just in case). There is no need to convert input files first.
fread running on any architecture will read a file from any architecture. Both \r and \n may be
embedded in character strings (including column names) provided the field is quoted.
Decimal separator and locale: fread(...,dec=",") should just work. fread uses C function
strtod to read numeric data; e.g., 1.23 or 1,23. strtod retrieves the decimal separator (. or ,
usually) from the locale of the R session rather than as an argument passed to the strtod func-
tion. So for fread(...,dec=",") to work, fread changes this (and only this) R session’s locale
temporarily to a locale which provides the desired decimal separator.
On Windows, "French_France.1252" is tried which should be available as standard (any locale with
comma decimal separator would suffice) and on unix "fr_FR.utf8" (you may need to install this
locale on unix). fread() is very careful to set the locale back again afterwards, even if the function
fails with an error. The choice of locale is determined by options()$datatable.fread.dec.locale.
46 fread
This may be a vector of locale names and if so they will be tried in turn until the desired dec is ob-
tained; thus allowing more than two different decimal separators to be selected. This is a new feature
in v1.9.6 and is experimental. In case of problems, turn it off with options(datatable.fread.dec.experiment=FALSE).
Quotes:
When quote is a single character,
• Spaces and other whitespace (other than sep and \n) may appear in unquoted character fields,
e.g., ...,2,Joe Bloggs,3.14,....
• When character columns are quoted, they must start and end with that quoting character
immediately followed by sep or \n, e.g., ...,2,"Joe Bloggs",3.14,....
In essence quoting character fields are required only if sep or \n appears in the string value.
Quoting may be used to signify that numeric data should be read as text. Unescaped quotes
may be present in a quoted field, e.g., ...,2,"Joe, "Bloggs"",3.14,..., as well as escaped
quotes, e.g., ...,2,"Joe \",Bloggs\"",3.14,....
If an embedded quote is followed by the separator inside a quoted field, the embedded quotes
up to that point in that field must be balanced; e.g. ...,2,"www.blah?x="one",y="two"",3.14,....
On those fields that do not satisfy these conditions, e.g., fields with unbalanced quotes, fread
re-attempts that field as if it isn’t quoted. This is quite useful in reading files that contains
fields with unbalanced quotes as well, automatically.
Value
A data.table by default. A data.frame when argument data.table=FALSE; e.g. options(datatable.fread.datatable
References
Background :
https://cran.r-project.org/doc/manuals/R-data.html
http://stackoverflow.com/questions/1727772/quickly-reading-very-large-tables-as-dataframes-in-r
http://www.biostat.jhsph.edu/~rpeng/docs/R-large-tables.html
http://www.cerebralmastication.com/2009/11/loading-big-data-into-r/
http://stackoverflow.com/questions/9061736/faster-than-scan-with-rcpp
http://stackoverflow.com/questions/415515/how-can-i-read-and-manipulate-csv-file-data-in-c
http://stackoverflow.com/questions/9352887/strategies-for-reading-in-csv-files-in-pieces
http://stackoverflow.com/questions/11782084/reading-in-large-text-files-in-r
http://stackoverflow.com/questions/45972/mmap-vs-reading-blocks
http://stackoverflow.com/questions/258091/when-should-i-use-mmap-for-file-access
fread 47
http://stackoverflow.com/a/9818473/403310
http://stackoverflow.com/questions/9608950/reading-huge-files-using-memory-mapped-files
finagler = "to get or achieve by guile or manipulation" http://dictionary.reference.com/
browse/finagler
See Also
read.csv, url, Sys.setlocale
Examples
## Not run:
# Demo speed-up
n = 1e6
DT = data.table( a=sample(1:1000,n,replace=TRUE),
b=sample(1:1000,n,replace=TRUE),
c=rnorm(n),
d=sample(c("foo","bar","baz","qux","quux"),n,replace=TRUE),
e=rnorm(n),
f=sample(1:1000,n,replace=TRUE) )
DT[2,b:=NA_integer_]
DT[4,c:=NA_real_]
DT[3,d:=NA_character_]
DT[5,d:=""]
DT[2,e:=+Inf]
DT[3,e:=-Inf]
write.table(DT,"test.csv",sep=",",row.names=FALSE,quote=FALSE)
cat("File size (MB):", round(file.info("test.csv")$size/1024^2),"\n")
# 50 MB (1e6 rows x 6 columns)
system.time(DF1 <-read.csv("test.csv",stringsAsFactors=FALSE))
# 60 sec (first time in fresh R session)
require(data.table)
if(all(sapply(c("sqldf", "ff"), requireNamespace, quietly = TRUE))) {
require(sqldf)
require(ff)
identical(DF1,DF2)
all.equal(as.data.table(DF1), DT)
identical(DF1,within(SQLDF,{b<-as.integer(b);c<-as.numeric(c)}))
identical(DF1,within(as.data.frame(FFDF),d<-as.character(d)))
}
# Scaling up ...
l = vector("list",10)
for (i in 1:10) l[[i]] = DT
DTbig = rbindlist(l)
tables()
write.table(DTbig,"testbig.csv",sep=",",row.names=FALSE,quote=FALSE)
# 500MB (10 million rows x 6 columns)
download.file("http://stat-computing.org/dataexpo/2009/2008.csv.bz2",
destfile="2008.csv.bz2")
# 109MB (compressed)
system("bunzip2 2008.csv.bz2")
# 658MB (7,009,728 rows x 29 columns)
colClasses = sapply(read.csv("2008.csv",nrows=100),class)
# 4 character, 24 integer, 1 logical. Incorrect.
colClasses = sapply(read.csv("2008.csv",nrows=200),class)
# 5 character, 24 integer. Correct. Might have missed data only using 100 rows
# since read.table assumes colClasses is correct.
table(sapply(DT,class))
# 5 character and 24 integer columns. Correct without needing to worry about colClasses
# issue above.
## End(Not run)
# Numerical precision :
DT = fread("A\n1.010203040506070809010203040506\n")
# TODO: add numerals=c("allow.loss", "warn.loss", "no.loss") from base::read.table, +"use.Rmpfr"
typeof(DT$A)=="double" # currently "allow.loss" with no option
# colClasses
data = "A,B,C,D\n1,3,5,7\n2,4,6,8\n"
fread(data, colClasses=c(B="character",C="character",D="character")) # as read.csv
fread(data, colClasses=list(character=c("B","C","D"))) # saves typing
fread(data, colClasses=list(character=2:4)) # same using column numbers
50 fsort
# drop
fread(data, colClasses=c("B"="NULL","C"="NULL")) # as read.csv
fread(data, colClasses=list(NULL=c("B","C"))) #
fread(data, drop=c("B","C")) # same but less typing, easier to read
fread(data, drop=2:3) # same using column numbers
# select
# (in read.csv you need to work out which to drop)
fread(data, select=c("A","D")) # less typing, easier to read
fread(data, select=c(1,4)) # same using column numbers
# check.names usage
fread("a b,a b\n1,2\n")
fread("a b,a b\n1,2\n", check.names=TRUE) # no duplicates + syntactically valid names
Description
Similar to base::sort but parallel. Experimental.
Usage
fsort(x, decreasing = FALSE, na.last = FALSE, internal=FALSE, verbose=FALSE, ...)
Arguments
x A vector. Type double, currently.
decreasing Decreasing order?
na.last Control treatment of NAs. If TRUE, missing values in the data are put last; if
FALSE, they are put first; if NA, they are removed; if "keep" they are kept with
rank NA.
internal Internal use only. Temporary variable. Will be removed.
verbose Print tracing information.
... Not sure yet. Should be consistent with base R.
fwrite 51
Details
Returns the input in sorted order. Fast using parallelism.
Value
The input in sorted order.
Examples
x = runif(1e6)
system.time(ans1 <- sort(x, method="quick"))
system.time(ans2 <- fsort(x))
identical(ans1, ans2)
Description
As write.csv but much faster (e.g. 2 seconds versus 1 minute) and just as flexible. Modern
machines almost surely have more than one CPU so fwrite uses them; on all operating systems
including Linux, Mac and Windows.
This is new functionality as of Nov 2016. We may need to refine argument names and defaults.
Usage
fwrite(x, file = "", append = FALSE, quote = "auto",
sep = ",", sep2 = c("","|",""),
eol = if (.Platform$OS.type=="windows") "\r\n" else "\n",
na = "", dec = ".", row.names = FALSE, col.names = TRUE,
qmethod = c("double","escape"),
logical01 = getOption("datatable.logical01", FALSE), # due to change to TRUE; see NEWS
logicalAsInt = logical01, # deprecated
dateTimeAs = c("ISO","squash","epoch","write.csv"),
buffMB = 8L, nThread = getDTthreads(),
showProgress = getOption("datatable.showProgress", interactive()),
verbose = getOption("datatable.verbose", FALSE))
Arguments
x Any list of same length vectors; e.g. data.frame and data.table.
file Output file name. "" indicates output to the console.
append If TRUE, the file is opened in append mode and column names (header row) are
not written.
52 fwrite
quote When "auto", character fields, factor fields and column names will only be sur-
rounded by double quotes when they need to be; i.e., when the field contains the
separator sep, a line ending \n, the double quote itself or (when list columns
are present) sep2[2] (see sep2 below). If FALSE the fields are not wrapped
with quotes even if this would break the CSV due to the contents of the field.
If TRUE double quotes are always included other than around numeric fields, as
write.csv.
sep The separator between columns. Default is ",".
sep2 For columns of type list where each item is an atomic vector, sep2 controls
how to separate items within the column. sep2[1] is written at the start of
the output field, sep2[2] is placed between each item and sep2[3] is written
at the end. sep2[1] and sep2[3] may be any length strings including empty
"" (default). sep2[2] must be a single character and (when list columns are
present and therefore sep2 is used) different from both sep and dec. The default
(|) is chosen to visually distinguish from the default sep. In speaking, writing
and in code comments we may refer to sep2[2] as simply "sep2".
eol Line separator. Default is "\r\n" for Windows and "\n" otherwise.
na The string to use for missing values in the data. Default is a blank string "".
dec The decimal separator, by default ".". See link in references. Cannot be the
same as sep.
row.names Should row names be written? For compatibility with data.frame and write.csv
since data.table never has row names. Hence default FALSE unlike write.csv.
col.names Should the column names (header row) be written? The default is TRUE. How-
ever, if missing, append=TRUE and the file already exists, the default is set to
FALSE for convenience to prevent column names appearing again mid file.
qmethod A character string specifying how to deal with embedded double quote charac-
ters when quoting strings.
• "escape" - the quote character (as well as the backslash character) is escaped
in C style by a backslash, or
• "double" (default, same as write.csv), in which case the double quote is
doubled with another one.
logical01 Should logical values be written as 1 and 0 rather than "TRUE" and "FALSE"?
logicalAsInt Deprecated. Old name for ‘logical01‘. Name change for consistency with
‘fread‘ for which ‘logicalAsInt‘ would not make sense.
dateTimeAs How Date/IDate, ITime and POSIXct items are written.
• "ISO" (default) - 2016-09-12, 18:12:16 and 2016-09-12T18:12:16.999999Z.
0, 3 or 6 digits of fractional seconds are printed if and when present for con-
venience, regardless of any R options such as digits.secs. The idea be-
ing that if milli and microseconds are present then you most likely want to
retain them. R’s internal UTC representation is written faithfully to encour-
age ISO standards, stymie timezone ambiguity and for speed. An option to
consider is to start R in the UTC timezone simply with "$ TZ='UTC' R"
at the shell (NB: it must be one or more spaces between TZ='UTC' and R,
anything else will be silently ignored; this TZ setting applies just to that R
process) or Sys.setenv(TZ='UTC') at the R prompt and then continue as
if UTC were local time.
fwrite 53
Details
fwrite began as a community contribution with pull request #1613 by Otto Seiskari. This gave Matt
Dowle the impetus to specialize the numeric formatting and to parallelize: http://blog.h2o.ai/
2016/04/fast-csv-writing-for-r/. Final items were tracked in issue #1664 such as automatic
quoting, bit64::integer64 support, decimal/scientific formatting exactly matching write.csv
between 2.225074e-308 and 1.797693e+308 to 15 significant figures, row.names, dates (between
0000-03-01 and 9999-12-31), times and sep2 for list columns where each cell can itself be a
vector.
54 fwrite
References
http://howardhinnant.github.io/date_algorithms.html
https://en.wikipedia.org/wiki/Decimal_mark
See Also
setDTthreads, fread, write.csv, write.table, bit64::integer64
Examples
DF = data.frame(A=1:3, B=c("foo","A,Name","baz"))
fwrite(DF)
write.csv(DF, row.names=FALSE, quote=FALSE) # same
DF = data.frame(A=c(2.1,-1.234e-307,pi), B=c("foo","A,Name","bar"))
fwrite(DF, quote='auto') # Just DF[2,2] is auto quoted
write.csv(DF, row.names=FALSE) # same numeric formatting
DT = data.table(A=c(2,5.6,-3),B=list(1:3,c("foo","A,Name","bar"),round(pi*1:3,2)))
fwrite(DT)
fwrite(DT, sep="|", sep2=c("{",",","}"))
## Not run:
set.seed(1)
DT = as.data.table( lapply(1:10, sample,
x=as.numeric(1:5e7), size=5e6)) # 382MB
system.time(fwrite(DT, "/dev/shm/tmp1.csv")) # 0.8s
system.time(write.csv(DT, "/dev/shm/tmp2.csv", # 60.6s
quote=FALSE, row.names=FALSE))
system("diff /dev/shm/tmp1.csv /dev/shm/tmp2.csv") # identical
set.seed(1)
N = 1e7
DT = data.table(
str1=sample(sprintf("%010d",sample(N,1e5,replace=TRUE)), N, replace=TRUE),
str2=sample(sprintf("%09d",sample(N,1e5,replace=TRUE)), N, replace=TRUE),
str3=sample(sapply(sample(2:30, 100, TRUE), function(n)
paste0(sample(LETTERS, n, TRUE), collapse="")), N, TRUE),
str4=sprintf("%05d",sample(sample(1e5,50),N,TRUE)),
num1=sample(round(rnorm(1e6,mean=6.5,sd=15),2), N, replace=TRUE),
num2=sample(round(rnorm(1e6,mean=6.5,sd=15),10), N, replace=TRUE),
str5=sample(c("Y","N"),N,TRUE),
str6=sample(c("M","F"),N,TRUE),
int1=sample(ceiling(rexp(1e6)), N, replace=TRUE),
int2=sample(N,N,replace=TRUE)-N/2
) # 774MB
system.time(fwrite(DT,"/dev/shm/tmp1.csv")) # 1.1s
groupingsets 55
system.time(write.csv(DT,"/dev/shm/tmp2.csv", # 63.2s
row.names=FALSE, quote=FALSE))
system("diff /dev/shm/tmp1.csv /dev/shm/tmp2.csv") # identical
unlink("/dev/shm/tmp1.csv")
unlink("/dev/shm/tmp2.csv")
## End(Not run)
Description
Calculate aggregates at various levels of groupings producing multiple (sub-)totals. Reflects SQLs
GROUPING SETS operations.
Usage
rollup(x, ...)
## S3 method for class 'data.table'
rollup(x, j, by, .SDcols, id = FALSE, ...)
cube(x, ...)
## S3 method for class 'data.table'
cube(x, j, by, .SDcols, id = FALSE, ...)
groupingsets(x, ...)
## S3 method for class 'data.table'
groupingsets(x, j, by, sets, .SDcols, id = FALSE, jj, ...)
Arguments
x data.table.
... argument passed to custom user methods. Ignored for data.table methods.
j expression passed to data.table j.
by character column names by which we are grouping.
sets list of character vector reflecting grouping sets, used in groupingsets for flex-
ibility.
.SDcols columns to be used in j expression in .SD object.
id logical default FALSE. If TRUE it will add leading column with bit mask of group-
ing sets.
jj quoted version of j argument, for convenience. When provided function will
ignore j argument.
56 groupingsets
Details
All three functions rollup, cube, groupingsets are generic methods, data.table methods are
provided.
Value
References
http://www.postgresql.org/docs/9.5/static/queries-table-expressions.html#QUERIES-GROUPING-SETS
http://www.postgresql.org/docs/9.5/static/functions-aggregate.html#FUNCTIONS-GROUPING-TABLE
See Also
data.table, rbindlist
Examples
n = 24L
set.seed(25)
DT <- data.table(
color = sample(c("green","yellow","red"), n, TRUE),
year = as.Date(sample(paste0(2011:2015,"-01-01"), n, TRUE)),
status = as.factor(sample(c("removed","active","inactive","archived"), n, TRUE)),
amount = sample(1:5, n, TRUE),
value = sample(c(3, 3.5, 2.5, 2), n, TRUE)
)
# rollup
rollup(DT, j = sum(value), by = c("color","year","status")) # default id=FALSE
rollup(DT, j = sum(value), by = c("color","year","status"), id=TRUE)
rollup(DT, j = lapply(.SD, sum), by = c("color","year","status"), id=TRUE, .SDcols="value")
rollup(DT, j = c(list(count=.N), lapply(.SD, sum)), by = c("color","year","status"), id=TRUE)
# cube
cube(DT, j = sum(value), by = c("color","year","status"), id=TRUE)
cube(DT, j = lapply(.SD, sum), by = c("color","year","status"), id=TRUE, .SDcols="value")
cube(DT, j = c(list(count=.N), lapply(.SD, sum)), by = c("color","year","status"), id=TRUE)
# groupingsets
groupingsets(DT, j = c(list(count=.N), lapply(.SD, sum)), by = c("color","year","status"),
sets = list("color", c("year","status"), character()), id=TRUE)
IDateTime 57
Description
Date and time classes with integer storage for fast sorting and grouping. Still experimental!
Usage
as.IDate(x, ...)
## Default S3 method:
as.IDate(x, ..., tz = attr(x, "tzone"))
## S3 method for class 'Date'
as.IDate(x, ...)
## S3 method for class 'IDate'
as.Date(x, ...)
## S3 method for class 'IDate'
as.POSIXct(x, tz = "UTC", time = 0, ...)
## S3 method for class 'IDate'
round(x, digits = c("weeks", "months", "quarters","years"), ...)
as.ITime(x, ...)
## Default S3 method:
as.ITime(x, ...)
## S3 method for class 'ITime'
as.POSIXct(x, tz = "UTC", date = as.Date(Sys.time()), ...)
## S3 method for class 'ITime'
as.character(x, ...)
## S3 method for class 'ITime'
format(x, ...)
IDateTime(x, ...)
## Default S3 method:
IDateTime(x, ...)
second(x)
minute(x)
hour(x)
yday(x)
wday(x)
mday(x)
week(x)
isoweek(x)
month(x)
quarter(x)
year(x)
58 IDateTime
Arguments
x an object
... arguments to be passed to or from other methods. For as.IDate.default, ar-
guments are passed to as.Date. For as.ITime.default, arguments are passed
to as.POSIXlt.
tz time zone (see strptime).
date date object convertable with as.IDate.
time time-of-day object convertable with as.ITime.
digits really units; one of the units listed for rounding. May be abbreviated.
Details
IDate is a date class derived from Date. It has the same internal representation as the Date class,
except the storage mode is integer. IDate is a relatively simple wrapper, and it should work in
almost all situations as a replacement for Date.
Functions that use Date objects generally work for IDate objects. This package provides specific
methods for IDate objects for mean, cut, seq, c, rep, and split to return an IDate object.
ITime is a time-of-day class stored as the integer number of seconds in the day. as.ITime does not
allow days longer than 24 hours. Because ITime is stored in seconds, you can add it to a POSIXct
object, but you should not add it to a Date object.
Conversions to and from Date and POSIXct formats are provided.
ITime does not account for time zones. When converting ITime and IDate to POSIXct with
as.POSIXct, a time zone may be specified.
In as.POSIXct methods for ITime and IDate, the second argument is required to be tz based
on the generic template, but to make converting easier, the second argument is interpreted as a date
instead of a time zone if it is of type IDate or ITime. Therefore, you can use either of the following:
as.POSIXct(time, date) or as.POSIXct(date,time).
IDateTime takes a date-time input and returns a data table with columns date and time.
Using integer storage allows dates and/or times to be used as data table keys. With positive integers
with a range less than 100,000, grouping and sorting is fast because radix sorting can be used (see
sort.list).
Several convenience functions like hour and quarter are provided to group or extract by hour,
month, and other date-time intervals. as.POSIXlt is also useful. For example, as.POSIXlt(x)$mon
is the integer month. The R base convenience functions weekdays, months, and quarters can
also be used, but these return character values, so they must be converted to factors for use with
data.table. isoweek is ISO 8601-consistent.
The round method for IDate’s is useful for grouping and plotting. It can round to weeks, months,
quarters, and years.
Value
For as.IDate, a class of IDate and Date with the date stored as the number of days since some
origin.
For as.ITime, a class of ITime stored as the number of seconds in the day.
IDateTime 59
For IDateTime, a data table with columns idate and itime in IDate and ITime format.
second, minute, hour, yday, wday, mday, week, month, quarter, and year return integer values
for second, minute, hour, day of year, day of week, day of month, week, month, quarter, and year,
respectively.
These values are all taken directly from the POSIXlt representation of x, with the notable difference
that while yday, wday, and mon are all 0-based, here they are 1-based.
Author(s)
Tom Short, t.short@ieee.org
References
G. Grothendieck and T. Petzoldt, “Date and Time Classes in R,” R News, vol. 4, no. 1, June 2004.
H. Wickham, http://gist.github.com/10238.
ISO 8601, http://www.iso.org/iso/home/standards/iso8601.htm
See Also
as.Date, as.POSIXct, strptime, DateTimeClasses
Examples
# create IDate:
(d <- as.IDate("2001-01-01"))
# create ITime:
(t <- as.ITime("10:45"))
(t <- as.ITime("10:45:04"))
as.POSIXct("2001-01-01") + as.ITime("10:45")
as.POSIXct(af$idate)
as.POSIXct(af$idate, time = af$itime)
as.POSIXct(af$idate, af$itime)
as.POSIXct(af$idate, time = af$itime, tz = "GMT")
as.POSIXct(af$itime, af$idate)
as.POSIXct(af$itime) # uses today's date
Description
Creates a data.table to be passed in as the i to a [.data.table join.
Usage
# DT[J(...)] # J() only for use inside DT[...].
SJ(...) # DT[SJ(...)]
CJ(..., sorted = TRUE, unique = FALSE) # DT[CJ(...)]
Arguments
... Each argument is a vector. Generally each vector is the same length but if they
are not then the usual silent repetition is applied.
sorted logical. Should the input order be retained?
unique logical. When TRUE, only unique values of each vectors are used (automatically).
Details
SJ and CJ are convenience functions for creating a data.table in the context of a data.table ’query’
on x.
x[data.table(id)] is the same as x[J(id)] but the latter is more readable. Identical alternatives
are x[list(id)] and x[.(id)].
x must have a key when passing in a join table as the i. See [.data.table
Value
J : the same result as calling list. J is a direct alias for list but results in clearer more readable
code.
SJ : (S)orted (J)oin. The same value as J() but additionally setkey() is called on all the columns
in the order they were passed in to SJ. For efficiency, to invoke a binary merge rather than a
repeated binary full search for each row of i.
last 61
CJ : (C)ross (J)oin. A data.table is formed from the cross product of the vectors. For example,
10 ids, and 100 dates, CJ returns a 1000 row table containing all the dates for all the ids. It
gains sorted, which by default is TRUE for backwards compatibility. FALSE retains input
order.
See Also
data.table, test.data.table
Examples
DT = data.table(A=5:1,B=letters[5:1])
setkey(DT,B) # re-orders table and marks it sorted.
DT[J("b")] # returns the 2nd row
DT[.("b")] # same. Style of package plyr.
DT[list("b")] # same
# CJ usage examples
CJ(c(5,NA,1), c(1,3,2)) # sorted and keyed data.table
do.call(CJ, list(c(5,NA,1), c(1,3,2))) # same as above
CJ(c(5,NA,1), c(1,3,2), sorted=FALSE) # same order as input, unkeyed
# use for 'unique=' argument
x = c(1,1,2)
y = c(4,6,4)
CJ(x, y, unique=TRUE) # unique(x) and unique(y) are computed automatically
Description
Returns the last item of a vector or list, or the last row of a data.frame or data.table.
Usage
last(x, ...)
Arguments
Value
If no other arguments are supplied it depends on the type of x. The last item of a vector or list. The
last row of a data.frame or data.table. Otherwise, whatever xts::last returns (if package xts
has been loaded, otherwise a helpful error).
If any argument is supplied in addition to x (such as n or keep in xts::last), regardless of x’s
type, then xts::last is called if xts has been loaded, otherwise a helpful error.
See Also
NROW, head, tail, first
Examples
last(1:5) # [1] 5
x = data.table(x=1:5, y=6:10)
last(x) # same as x[5]
Description
Intended for use in i in [.data.table.
Usage
like(vector,pattern)
vector %like% pattern
Arguments
vector Either a character vector or a factor. A factor is faster.
pattern Passed on to grepl.
Value
Logical vector, TRUE for items that match pattern.
Note
Current implementation does not make use of sorted keys.
See Also
data.table, grepl
melt.data.table 63
Examples
DT = data.table(Name=c("Mary","George","Martha"), Salary=c(2,3,4))
DT[Name %like% "^Mar"]
Description
An S3 method for melting data.tables written in C for speed and memory efficiency. Since
v1.9.6, melt.data.table allows melting into multiple columns simultaneously.
It is not necessary to load reshape2 any more. But if you have to, then load reshape2 package
before loading data.table.
Usage
## fast melt a data.table
## S3 method for class 'data.table'
melt(data, id.vars, measure.vars,
variable.name = "variable", value.name = "value",
..., na.rm = FALSE, variable.factor = TRUE,
value.factor = FALSE,
verbose = getOption("datatable.verbose"))
Arguments
data A data.table object to melt.
id.vars vector of id variables. Can be integer (corresponding id column numbers) or
character (id column names) vector. If missing, all non-measure columns will
be assigned to it. If integer, must be positive; see Details.
measure.vars Measure variables for melting. Can be missing, vector, list, or pattern-based.
• When missing, measure.vars will become all columns outside id.vars.
• Vector can be integer (implying column numbers) or character (column
names).
• list is a generalization of the vector version – each element of the list
(which should be integer or character as above) will become a melted
column.
• Pattern-based column matching can be achieved with the regular expression-
based patterns syntax; multiple patterns will produce multiple columns.
For convenience/clarity in the case of multiple melted columns, resulting col-
umn names can be supplied as names to the elements measure.vars (in the
list and patterns usages). See also Examples.
variable.name name for the measured variable names column. The default name is 'variable'.
64 melt.data.table
value.name name for the molten data values column(s). The default name is 'value'. Mul-
tiple names can be provided here for the case when measure.vars is a list,
though note well that the names provided in measure.vars take precedence.
na.rm If TRUE, NA values will be removed from the molten data.
variable.factor
If TRUE, the variable column will be converted to factor, else it will be a
character column.
value.factor If TRUE, the value column will be converted to factor, else the molten value
type is left unchanged.
verbose TRUE turns on status and information messages to the console. Turn this on by
default using options(datatable.verbose=TRUE). The quantity and types of
verbosity may be expanded in future.
... any other arguments to be passed to/from other methods.
Details
If id.vars and measure.vars are both missing, all non-numeric/integer/logical columns are
assigned as id variables and the rest as measure variables. If only one of id.vars or measure.vars
is supplied, the rest of the columns will be assigned to the other. Both id.vars and measure.vars
can have the same column more than once and the same column can be both as id and measure
variables.
melt.data.table also accepts list columns for both id and measure variables.
When all measure.vars are not of the same type, they’ll be coerced according to the hierarchy list
> character > numeric > integer > logical. For example, if any of the measure variables is
a list, then entire value column will be coerced to a list. Note that, if the type of value column is
a list, na.rm = TRUE will have no effect.
From version 1.9.6, melt gains a feature with measure.vars accepting a list of character or
integer vectors as well to melt into multiple columns in a single function call efficiently. The
function patterns can be used to provide regular expression patterns. When used along with
melt, if cols argument is not provided, the patterns will be matched against names(data), for
convenience.
Attributes are preserved if all value columns are of the same type. By default, if any of the
columns to be melted are of type factor, it’ll be coerced to character type. This is to be com-
patible with reshape2’s melt.data.frame. To get a factor column, set value.factor = TRUE.
melt.data.table also preserves ordered factors.
Note that, as opposed to the (undocumented) behaviour of reshape2::melt, id.vars, when spec-
ified as numbers, must be between 1 and ncol(data).
Value
An unkeyed data.table containing the molten data.
See Also
dcast, https://cran.r-project.org/package=reshape
melt.data.table 65
Examples
set.seed(45)
require(data.table)
DT <- data.table(
i_1 = c(1:5, NA),
i_2 = c(NA,6,7,8,9,10),
f_1 = factor(sample(c(letters[1:3], NA), 6, TRUE)),
f_2 = factor(c("z", "a", "x", "c", "x", "x"), ordered=TRUE),
c_1 = sample(c(letters[1:3], NA), 6, TRUE),
d_1 = as.Date(c(1:3,NA,4:5), origin="2013-09-01"),
d_2 = as.Date(6:1, origin="2012-01-01"))
# add a couple of list cols
DT[, l_1 := DT[, list(c=list(rep(i_1, sample(5,1)))), by = i_1]$c]
DT[, l_2 := DT[, list(c=list(rep(c_1, sample(5,1)))), by = i_1]$c]
# on list
melt(DT, id=1, measure=c("l_1", "l_2")) # value is a list
melt(DT, id=1, measure=c("c_1", "l_1")) # c1 coerced to list
# on character
melt(DT, id=1, measure=c("c_1", "f_1")) # value is char
melt(DT, id=1, measure=c("c_1", "i_2")) # i2 coerced to char
# return 'NA' for missing columns, 'na.rm=TRUE' ignored due to list column
melt(DT, id=1:2, measure=patterns("l_", "c_"), na.rm=TRUE)
66 merge
Description
Fast merge of two data.tables. The data.table method behaves very similarly to that of data.frames
except that, by default, it attempts to merge
• at first based on the shared key columns, and if there are none,
• then based on key columns of the first argument x, and if there are none,
• then based on the common columns between the two data.tables.
Set the by, or by.x and by.y arguments explicitly to override this default.
Usage
## S3 method for class 'data.table'
merge(x, y, by = NULL, by.x = NULL, by.y = NULL, all = FALSE,
all.x = all, all.y = all, sort = TRUE, suffixes = c(".x", ".y"), no.dups = TRUE,
allow.cartesian=getOption("datatable.allow.cartesian"), # default FALSE
...)
Arguments
x, y data tables. y is coerced to a data.table if it isn’t one already.
by A vector of shared column names in x and y to merge on. This defaults to
the shared key columns between the two tables. If y has no key columns, this
defaults to the key of x.
by.x, by.y Vectors of column names in x and y to merge on.
all logical; all = TRUE is shorthand to save setting both all.x = TRUE and
all.y = TRUE.
all.x logical; if TRUE, then extra rows will be added to the output, one for each row in
x that has no matching row in y. These rows will have ’NA’s in those columns
that are usually filled with values from y. The default is FALSE, so that only rows
with data from both x and y are included in the output.
all.y logical; analogous to all.x above.
sort logical. If TRUE (default), the merged data.table is sorted by setting the key to
the by / by.x columns. If FALSE, the result is not sorted.
suffixes A character(2) specifying the suffixes to be used for making non-by col-
umn names unique. The suffix behaviour works in a similar fashion as the
merge.data.frame method does.
no.dups logical indicating that suffixes are also appended to non-by.y column names
in y when they have the same column name as any by.x.
allow.cartesian
See allow.cartesian in [.data.table.
... Not used at this time.
merge 67
Details
merge is a generic function in base R. It dispatches to either the merge.data.frame method or
merge.data.table method depending on the class of its first argument. Note that, unlike SQL, NA
is matched against NA (and NaN against NaN) while merging.
In versions <= v1.9.4, if the specified columns in by were not the key (or head of the key) of x or
y, then a copy is first re-keyed prior to performing the merge. This was less performant as well as
memory inefficient. The concept of secondary keys (implemented in v1.9.4) was used to overcome
this limitation from v1.9.6+. No deep copies are made any more, thereby improving performance
and memory efficiency. Also, there is better control for providing the columns to merge on with the
help of the newly implemented by.x and by.y arguments.
For a more data.table-centric way of merging two data.tables, see [.data.table; e.g., x[y, ...].
See FAQ 1.12 for a detailed comparison of merge and x[y, ...].
If any column names provided to by.x also occur in names(y) but not in by.y, then this data.table
method will add the suffixes to those column names. As of R v3.4.3, the data.frame method
will not (leading to duplicate column names in the result) but a patch has been proposed (see r-devel
thread here) which is looking likely to be accepted for a future version of R.
Value
A new data.table based on the merged data tables, and sorted by the columns set (or inferred
for) the by argument if argument sort is set to TRUE.
See Also
data.table, as.data.table, [.data.table, merge.data.frame
Examples
(dt1 <- data.table(A = letters[1:10], X = 1:10, key = "A"))
(dt2 <- data.table(A = letters[5:14], Y = 1:10, key = "A"))
merge(dt1, dt2)
merge(dt1, dt2, all = TRUE)
(dt1 <- data.table(A = c(rep(1L, 5), 2L), B = letters[rep(1:3, 2)], X = 1:6, key = "A,B"))
(dt2 <- data.table(A = c(rep(1L, 5), 2L), B = letters[rep(2:4, 2)], Y = 6:1, key = "A,B"))
merge(dt1, dt2)
merge(dt1, dt2, by="B", allow.cartesian=TRUE)
# test it more:
d1 <- data.table(a=rep(1:2,each=3), b=1:6, key="a,b")
d2 <- data.table(a=0:1, bb=10:11, key="a")
d3 <- data.table(a=0:1, key="a")
d4 <- data.table(a=0:1, b=0:1, key="a,b")
merge(d1, d2)
68 na.omit.data.table
merge(d2, d1)
merge(d1, d2, all=TRUE)
merge(d2, d1, all=TRUE)
merge(d3, d1)
merge(d1, d3)
merge(d1, d3, all=TRUE)
merge(d3, d1, all=TRUE)
merge(d1, d4)
merge(d1, d4, by="a", suffixes=c(".d1", ".d4"))
merge(d4, d1)
merge(d1, d4, all=TRUE)
merge(d4, d1, all=TRUE)
Description
This is a data.table method for the S3 generic stats::na.omit. The internals are written in C
for speed. See examples for benchmark timings.
bit64::integer64 type is also supported.
Usage
## S3 method for class 'data.table'
na.omit(object, cols=seq_along(object), invert=FALSE, ...)
Arguments
object A data.table.
cols A vector of column names (or numbers) on which to check for missing values.
Default is all the columns.
invert logical. If FALSE omits all rows with any missing values (default). TRUE returns
just those rows with missing values instead.
... Further arguments special methods could require.
patterns 69
Details
The data.table method consists of an additional argument cols, which when specified looks for
missing values in just those columns specified. The default value for cols is all the columns, to be
consistent with the default behaviour of stats::na.omit.
It does not add the attribute na.action as stats::na.omit does.
Value
A data.table with just the rows where the specified columns have no missing value in any of them.
See Also
data.table
Examples
DT = data.table(x=c(1,NaN,NA,3), y=c(NA_integer_, 1:3), z=c("a", NA_character_, "b", "c"))
# default behaviour
na.omit(DT)
# omit rows where 'x' has a missing value
na.omit(DT, cols="x")
# omit rows where either 'x' or 'y' have missing values
na.omit(DT, cols=c("x", "y"))
## Not run:
# Timings on relatively large data
set.seed(1L)
DT = data.table(x = sample(c(1:100, NA_integer_), 5e7L, TRUE),
y = sample(c(rnorm(100), NA), 5e7L, TRUE))
system.time(ans1 <- na.omit(DT)) ## 2.6 seconds
system.time(ans2 <- stats:::na.omit.data.frame(DT)) ## 29 seconds
# identical? check each column separately, as ans2 will have additional attribute
all(sapply(1:2, function(i) identical(ans1[[i]], ans2[[i]]))) ## TRUE
## End(Not run)
Description
patterns returns the matching indices in the argument cols corresponding to the regular expres-
sion patterns provided. The patterns must be supported by grep.
From v1.9.6, melt.data.table has an enhanced functionality in which measure.vars argument
can accept a list of column names and melt them into separate columns. See the Efficient reshaping using
data.tables vignette linked below to learn more.
70 print.data.table
Usage
patterns(..., cols=character(0))
Arguments
See Also
melt, https://github.com/Rdatatable/data.table/wiki/Getting-started
Examples
DT = data.table(x1 = 1:5, x2 = 6:10, y1 = letters[1:5], y2 = letters[6:10])
# melt all columns that begin with 'x' & 'y', respectively, into separate columns
melt(DT, measure.vars = patterns("^x", "^y", cols=names(DT)))
# when used with melt, 'cols' is implictly assumed to be names of input
# data.table, if not provided.
melt(DT, measure.vars = patterns("^x", "^y"))
Description
Usage
Arguments
x A data.table.
topn The number of rows to be printed from the beginning and end of tables with
more than nrows rows.
nrows The number of rows which will be printed before truncation is enforced.
class If TRUE, the resulting output will include above each column its storage class (or
a self-evident abbreviation thereof).
row.names If TRUE, row indices will be printed alongside x.
col.names One of three flavours for controlling the display of column names in output.
"auto" includes column names above the data, as well as below the table if
nrow(x) > 20. "top" excludes this lower register when applicable, and "none"
suppresses column names altogether (as well as column classes if class = TRUE.
print.keys If TRUE, any key and/or index currently assigned to x will be printed prior to
the preview of the data.
quote If TRUE, all output will appear in quotes, as in print.default.
... Other arguments ultimately passed to format.
Details
By default, with an eye to the typically large number of observations in a codedata.table, only the
beginning and end of the object are displayed (specifically, head(x, topn) and tail(x, topn)
are displayed unless nrow(x) < nrows, in which case all rows will print).
See Also
print.default
Examples
#output compression
DT <- data.table(a = 1:1000)
print(DT, nrows = 100, topn = 4)
Description
Same as do.call("rbind", l) on data.frames, but much faster. See DETAILS for more.
Usage
rbindlist(l, use.names=fill, fill=FALSE, idcol=NULL)
# rbind(..., use.names=TRUE, fill=FALSE, idcol=NULL)
Arguments
l A list containing data.table, data.frame or list objects. At least one of the
inputs should have column names set. ... is the same but you pass the objects
by name separately.
use.names If TRUE items will be bound by matching column names. By default FALSE for
rbindlist (for backwards compatibility) and TRUE for rbind (consistency with
base). Columns with duplicate names are bound in the order of occurrence,
similar to base. When TRUE, at least one item of the input list has to have
non-null column names.
fill If TRUE fills missing columns with NAs. By default FALSE. When TRUE, use.names
has to be TRUE, and all items of the input list has to have non-null column names.
idcol Generates an index column. Default (NULL) is not to. If idcol=TRUE then the
column is auto named .id. Alternatively the column name can be directly pro-
vided, e.g., idcol = "id".
If input is a named list, ids are generated using them, else using integer vector
from 1 to length of input list. See examples.
Details
Each item of l can be a data.table, data.frame or list, including NULL (skipped) or an empty
object (0 rows). rbindlist is most useful when there are a variable number of (potentially many)
objects to stack, such as returned by lapply(fileNames, fread). rbind however is most useful to
stack two or three objects which you know in advance. ... should contain at least one data.table
for rbind(...) to call the fast method and return a data.table, whereas rbindlist(l) always
returns a data.table even when stacking a plain list with a data.frame, for example.
In versions <= v1.9.2, each item for rbindlist should have the same number of columns as the
first non empty item. rbind.data.table gained a fill argument to fill missing columns with NA
in v1.9.2, which allowed for rbind(...) binding unequal number of columns.
rbindlist 73
In version > v1.9.2, these functionalities were extended to rbindlist (and written entirely in C
for speed). rbindlist has use.names argument, which is set to FALSE by default for backwards
compatibility. It also contains fill argument as well and can bind unequal columns when set to
TRUE.
With these changes, the only difference between rbind(...) and rbindlist(l) is their default
argument use.names.
If column i of input items do not all have the same type; e.g, a data.table may be bound with a
list or a column is factor while others are character types, they are coerced to the highest type
(SEXPTYPE).
Note that any additional attributes that might exist on individual items of the input list would not be
preserved in the result.
Value
See Also
data.table, split.data.table
Examples
# default case
DT1 = data.table(A=1:3,B=letters[1:3])
DT2 = data.table(A=4:5,B=letters[4:5])
l = list(DT1,DT2)
rbindlist(l)
Description
A convenience function for generating a run-length type id column to be used in grouping opera-
tions. It accepts atomic vectors, lists, data.frames or data.tables as input.
Usage
rleid(..., prefix=NULL)
rleidv(x, cols=seq_along(x), prefix=NULL)
Arguments
Details
At times aggregation (or grouping) operations need to be performed where consecutive runs of
identical values should belong to the same group (See rle). The use for such a function has come
up repeatedly on StackOverflow, see the See Also section. This function allows to generate "run-
length" groups directly.
rleid is designed for interactive use and accepts a sequence of vectors as arguments. For program-
ming, rleidv might be more useful.
Value
When prefix = NULL, an integer vector with same length as NROW(x), else a character vector with
the value in prefix prefixed to the ids obtained.
See Also
Examples
DT = data.table(grp=rep(c("A", "B", "C", "A", "B"), c(2,2,3,1,2)), value=1:10)
rleid(DT$grp) # get run-length ids
rleidv(DT, "grp") # same as above
Description
Convenience functions for generating a unique row ids within each group. It accepts atomic vectors,
lists, data.frames or data.tables as input.
rowid is intended for interactive use, particularly along with the function dcast to generate unique
ids directly in the formula.
rowidv(DT, cols=c("x", "y")) is equivalent to column N in the code DT[, N := seq_len(.N), by=c("x", "y")].
See examples for more.
Usage
rowid(..., prefix=NULL)
rowidv(x, cols=seq_along(x), prefix=NULL)
Arguments
x A vector, list, data.frame or data.table.
... A sequence of numeric, integer64, character or logical vectors, all of same
length. For interactive use.
cols Only meaningful for lists, data.frames or data.tables. A character vector of col-
umn names (or numbers) of x.
prefix Either NULL (default) or a character vector of length=1 which is prefixed to the
row ids, returning a character vector (instead of an integer vector).
Value
When prefix = NULL, an integer vector with same length as NROW(x), else a character vector with
the value in prefix prefixed to the ids obtained.
See Also
dcast.data.table, rleid
76 setattr
Examples
DT = data.table(x=c(20,10,10,30,30,20), y=c("a", "a", "a", "b", "b", "b"), z=1:6)
rowid(DT$x) # 1,1,2,1,2,2
rowidv(DT, cols="x") # same as above
Description
In data.table, all set* functions change their input by reference. That is, no copy is made at all,
other than temporary working memory which is as large as one column. The only other data.table
operator that modifies input by reference is :=. Check out the See Also section below for other
set* function that data.table provides.
Usage
setattr(x,name,value)
setnames(x,old,new)
Arguments
x setnames accepts data.frame and data.table. setattr accepts any input;
e.g, list, columns of a data.frame or data.table.
name The character attribute name.
value The value to assign to the attribute or NULL removes the attribute, if present.
old When new is provided, character names or numeric positions of column names
to change. When new is not provided, the new column names, which must be
the same length as the number of columns. See examples.
new Optional. New column names, must be the same length as columns provided to
old argument.
setattr 77
Details
setnames operates on data.table and data.frame not other types like list and vector. It can
be used to change names by name with built-in checks and warnings (e.g., if any old names are
missing or appear more than once).
setattr is a more general function that allows setting of any attribute to an object by reference.
A very welcome change in R 3.1+ was that ‘names<-‘ and ‘colnames<-‘ no longer copy the entire
object as they used to (up to 4 times), see examples below. They now take a shallow copy. The ‘set*‘
functions in data.table are still useful because they don’t even take a shallow copy. This allows
changing names and attributes of a (usually very large) data.table in the global environment from
within functions. Like a database.
Value
The input is modified by reference, and returned (invisibly) so it can be used in compound state-
ments; e.g., setnames(DT,"V1", "Y")[, .N, by=Y]. If you require a copy, take a copy first
(using DT2=copy(DT)). See ?copy.
Note that setattr is also in package bit. Both packages merely expose R’s internal setAttrib
function at C level but differ in return value. bit::setattr returns NULL (invisibly) to remind you
the function is used for its side effect. data.table::setattr returns the changed object (invisibly)
for use in compound statements.
See Also
data.table, setkey, setorder, setcolorder, set, :=, setDT, setDF, copy
Examples
DT = data.table(a=1:3, b=4:6)
f = function(...) {
78 setcolorder
# ...
setattr(DT,"myFlag",TRUE) # by reference
# ...
localDT = copy(DT)
setattr(localDT,"myFlag2",TRUE)
# ...
invisible()
}
f()
attr(DT,"myFlag") # TRUE
attr(DT,"myFlag2") # NULL
Description
In data.table parlance, all set* functions change their input by reference. That is, no copy is
made at all, other than temporary working memory, which is as large as one column. The only
other data.table operator that modifies input by reference is :=. Check out the See Also section
below for other set* function data.table provides.
setcolorder reorders the columns of data.table, by reference, to the new order provided.
Usage
setcolorder(x, neworder)
Arguments
x A data.table.
neworder Character vector of the new column name ordering. May also be column num-
bers. If length(neworder) < length(x), the specified columns are moved in
order to the "front" of x.
Details
To reorder data.table columns, the idiomatic way is to use setcolorder(x, neworder), instead
of doing x <- x[, neworder, with=FALSE]. This is because the latter makes an entire copy of
the data.table, which maybe unnecessary in most situations. setcolorder also allows column
numbers instead of names for neworder argument, although we recommend using names as a good
programming practice.
Value
The input is modified by reference, and returned (invisibly) so it can be used in compound state-
ments. If you require a copy, take a copy first (using DT2 = copy(DT)). See ?copy.
setDF 79
See Also
setkey, setorder, setattr, setnames, set, :=, setDT, setDF, copy, getNumericRounding,
setNumericRounding
Examples
set.seed(45L)
DT = data.table(A=sample(3, 10, TRUE),
B=sample(letters[1:3], 10, TRUE), C=sample(10))
#incomplete specification
setcolorder(DT, "A")
Description
In data.table parlance, all set* functions change their input by reference. That is, no copy is
made at all, other than temporary working memory, which is as large as one column. The only
other data.table operator that modifies input by reference is :=. Check out the See Also section
below for other set* function data.table provides.
A helper function to convert a data.table or list of equal length to data.frame by reference.
Usage
setDF(x, rownames=NULL)
Arguments
x A data.table, data.frame or list of equal length.
rownames A character vector to assign as the row names of x.
Details
All data.table attributes including any keys of the input data.table are stripped off.
When using rownames, recall that the row names of a data.frame must be unique. By default, the
assigned set of row names is simply the sequence 1, ..., nrow(x) (or length(x) for lists).
Value
The input data.table is modified by reference to a data.frame and returned (invisibly). If you
require a copy, take a copy first (using DT2 = copy(DT)). See ?copy.
80 setDT
See Also
data.table, as.data.table, setDT, copy, setkey, setcolorder, setattr, setnames, set, :=,
setorder
Examples
X = data.table(x=1:5, y=6:10)
## convert 'X' to data.frame, without any copy.
setDF(X)
X = data.table(x=1:5, y=6:10)
## idem, assigning row names
setDF(X, rownames = LETTERS[1:5])
X = list(x=1:5, y=6:10)
# X is converted to a data.frame without any copy.
setDF(X)
Description
In data.table parlance, all set* functions change their input by reference. That is, no copy is
made at all, other than temporary working memory, which is as large as one column.. The only
other data.table operator that modifies input by reference is :=. Check out the See Also section
below for other set* function data.table provides.
setDT converts lists (both named and unnamed) and data.frames to data.tables by reference. This
feature was requested on Stackoverflow.
Usage
Arguments
Details
When working on large lists or data.frames, it might be both time and memory consuming to
convert them to a data.table using as.data.table(.), as this will make a complete copy of the
input object before to convert it to a data.table. The setDT function takes care of this issue by
allowing to convert lists - both named and unnamed lists and data.frames by reference instead.
That is, the input object is modified in place, no copy is being made.
Value
The input is modified by reference, and returned (invisibly) so it can be used in compound state-
ments; e.g., setDT(X)[, sum(B), by=A]. If you require a copy, take a copy first (using DT2 = copy(DT)).
See ?copy.
See Also
data.table, as.data.table, setDF, copy, setkey, setcolorder, setattr, setnames, set, :=,
setorder
Examples
set.seed(45L)
X = data.frame(A=sample(3, 10, TRUE),
B=sample(letters[1:3], 10, TRUE),
C=sample(10), stringsAsFactors=FALSE)
# setkey directly
X = list(a = 4:1, b=runif(4))
setDT(X, key="a")[]
# check.names argument
X = list(a=1:5, a=6:10)
setDT(X, check.names=TRUE)[]
82 setDTthreads
Description
Set and get number of threads to be used in data.table functions that are parallelized with
OpenMP. Default value 0 means to utilize all CPU available with an appropriate number of threads
calculated by OpenMP. getDTthreads() returns the number of threads that will be used. This af-
fects data.table only and does not change R itself or other packages using OpenMP. The most
common usage expected is setDTthreads(1) to limit data.table to one thread for pre-existing
explicitly parallel user code; e.g. via packages parallel and foreach. Otherwise, nested parallelism
may bite. As data.table becomes more parallel automatically internally, we expect explicit user
parallelism to be needed less often.
Attempting to setDTthreads() to more than the number of logical CPUs is intended to be ineffec-
tive; i.e., getDTthreads() will still return the number of logical CPUs in that case. Further, there is
a hard coded limit of 1024 threads (with warning when imposed) to prevent accidentally picking up
the value of INT_MAX (2 billion; i.e. unlimited) from omp_get_thread_limit(). We have followed
the advice of section 1.2.1.1 in the R-exts manual: "... or, better, for the regions in your code as
part of their specification... num_threads(nthreads).. That way you only control your own code and
not that of other OpenMP users." All the parallel region in data.table contain this directive. This is
mandated by a grep in the package’s quality control release procedure script.
Usage
setDTthreads(threads)
getDTthreads(verbose = getOption("datatable.verbose", FALSE))
Arguments
threads An integer >= 0. Default 0 means use all CPU available and leave the operating
system to multi task.
verbose Display the value returned by some OpenMP function calls.
Value
A length 1 integer. The old value is returned by setDTthreads so you can store that value and
pass it to setDTthreads again after the section of your code where you, probably, limited to one
thread.
setkey 83
Description
In data.table parlance, all set* functions change their input by reference. That is, no copy is
made at all, other than temporary working memory, which is as large as one column. The only
other data.table operator that modifies input by reference is :=. Check out the See Also section
below for other set* function data.table provides.
setkey() sorts a data.table and marks it as sorted (with an attribute sorted). The sorted columns
are the key. The key can be any columns in any order. The columns are sorted in ascending order
always. The table is changed by reference and is therefore very memory efficient.
setindex() creates an index (or indices) on provided columns. This index is simply an order of the
dataset’s according to the provided columns. This order is stored as a data.table attribute, and the
dataset retains the original order in memory. See the Secondary indices and auto indexing vignette
for more details.
key() returns the data.table’s key if it exists, and NULL if none exist.
haskey() returns a logical TRUE/FALSE depending on whether the data.table has a key (or not).
Usage
setkey(x, ..., verbose=getOption("datatable.verbose"), physical = TRUE)
setkeyv(x, cols, verbose=getOption("datatable.verbose"), physical = TRUE)
setindex(...)
setindexv(x, cols, verbose=getOption("datatable.verbose"))
key(x)
indices(x, vectors = FALSE)
haskey(x)
key(x) <- value # DEPRECATED, please use setkey or setkeyv instead.
Arguments
x A data.table.
... The columns to sort by. Do not quote the column names. If ... is missing (i.e.
setkey(DT)), all the columns are used. NULL removes the key.
cols A character vector of column names. For setindexv, this can be a list of
character vectors, in which case each element will be applied as an index.
value In (deprecated) key<-, a character vector (only) of column names.
verbose Output status and information.
physical TRUE changes the order of the data in RAM. FALSE adds a secondary key
a.k.a. index.
vectors logical scalar default FALSE, when set to TRUE then list of character vectors is
returned, each vector refers to one index.
84 setkey
Details
setkey reorders (or sorts) the rows of a data.table by the columns provided. In versions 1.9+,
for integer columns, a modified version of base’s counting sort is implemented, which allows
negative values as well. It is extremely fast, but is limited by the range of integer values being <=
1e5. If that fails, it falls back to a (fast) 4-pass radix sort for integers, implemented based on Pierre
Terdiman’s and Michael Herf’s code (see links below). Similarly, a very fast 6-pass radix order for
columns of type double is also implemented. This gives a speed-up of about 5-8x compared to
1.8.10 on setkey and all internal order/sort operations. Fast radix sorting is also implemented
for character and bit64::integer64 types.
The sort is stable; i.e., the order of ties (if any) is preserved, in both versions - <=1.8.10 and
>= 1.9.0.
In data.table versions <= 1.8.10, for columns of type integer, the sort is attempted with the
very fast "radix" method in sort.list. If that fails, the sort reverts to the default method in
order. For character vectors, data.table takes advantage of R’s internal global string cache and
implements a very efficient order, also exported as chorder.
In v1.7.8, the key<- syntax was deprecated. The <- method copies the whole table and we know of
no way to avoid that copy without a change in R itself. Please use the set* functions instead, which
make no copy at all. setkey accepts unquoted column names for convenience, whilst setkeyv
accepts one vector of column names.
The problem (for data.table) with the copy by key<- (other than being slower) is that R doesn’t
maintain the over allocated truelength, but it looks as though it has. Adding a column by ref-
erence using := after a key<- was therefore a memory overwrite and eventually a segfault; the
over allocated memory wasn’t really there after key<-’s copy. data.tables now have an attribute
.internal.selfref to catch and warn about such copies. This attribute has been implemented in
a way that is friendly with identical() and object.size().
For the same reason, please use the other set* functions which modify objects by reference, rather
than using the <- operator which results in copying the entire object.
It isn’t good programming practice, in general, to use column numbers rather than names. This is
why setkey and setkeyv only accept column names. If you use column numbers then bugs (possi-
bly silent) can more easily creep into your code as time progresses if changes are made elsewhere in
your code; e.g., if you add, remove or reorder columns in a few months time, a setkey by column
number will then refer to a different column, possibly returning incorrect results with no warning.
(A similar concept exists in SQL, where "select * from ..." is considered poor programming
style when a robust, maintainable system is required.) If you really wish to use column numbers,
it’s possible but deliberately a little harder; e.g., setkeyv(DT,colnames(DT)[1:2]).
Value
The input is modified by reference, and returned (invisibly) so it can be used in compound state-
ments; e.g., setkey(DT,a)[J("foo")]. If you require a copy, take a copy first (using DT2=copy(DT)).
copy() may also sometimes be useful before := is used to subassign to a column by reference. See
?copy.
Note
Despite its name, base::sort.list(x,method="radix") actually invokes a counting sort in R,
not a radix sort. See do_radixsort in src/main/sort.c. A counting sort, however, is particularly
setkey 85
suitable for sorting integers and factors, and we like it. In fact we like it so much that data.table
contains a counting sort algorithm for character vectors using R’s internal global string cache. This
is particularly fast for character vectors containing many duplicates, such as grouped data in a key
column. This means that character is often preferred to factor. Factors are still fully supported, in
particular ordered factors (where the levels are not in alphabetic order).
References
http://en.wikipedia.org/wiki/Radix_sort
http://en.wikipedia.org/wiki/Counting_sort
http://cran.at.r-project.org/web/packages/bit/index.html
http://stereopsis.com/radix.html
See Also
data.table, tables, J, sort.list, copy, setDT, setDF, set :=, setorder, setcolorder, setattr,
setnames, chorder, setNumericRounding
Examples
# Type 'example(setkey)' to run these at prompt and browse output
DT = data.table(A=5:1,B=letters[5:1])
DT # before
setkey(DT,B) # re-orders table and marks it sorted.
DT # after
tables() # KEY column reports the key'd columns
key(DT)
keycols = c("A","B")
setkeyv(DT,keycols) # rather than key(DT)<-keycols (which copies entire table)
DT = data.table(A=5:1,B=letters[5:1])
DT2 = DT # does not copy
setkey(DT2,B) # does not copy-on-write to DT2
identical(DT,DT2) # TRUE. DT and DT2 are two names for the same keyed table
DT = data.table(A=5:1,B=letters[5:1])
DT2 = copy(DT) # explicit copy() needed to copy a data.table
setkey(DT2,B) # now just changes DT2
identical(DT,DT2) # FALSE. DT and DT2 are now different tables
DT = data.table(A=5:1,B=letters[5:1])
setindex(DT) # set indices
setindex(DT, A)
setindex(DT, B)
indices(DT) # get indices single vector
indices(DT, vectors = TRUE) # get indices list
86 setNumericRounding
Description
Change rounding to 0, 1 or 2 bytes when joining, grouping or ordering numeric (i.e. double,
POSIXct) columns.
Usage
setNumericRounding(x)
getNumericRounding()
Arguments
Details
Computers cannot represent some floating point numbers (such as 0.6) precisely, using base 2. This
leads to unexpected behaviour when joining or grouping columns of type ’numeric’; i.e. ’double’,
see example below. In cases where this is undesirable, data.table allows rounding such data up to
approximately 11 s.f. which is plenty of digits for many cases. This is achieved by rounding the
last 2 bytes off the significand. Other possible values are 1 byte rounding, or no rounding (full
precision, default).
It’s bytes rather than bits because it’s tied in with the radix sort algorithm for sorting numerics which
sorts byte by byte. With the default rounding of 0 bytes, at most 8 passes are needed. With rounding
of 2 bytes, at most 6 passes are needed (and therefore might be a tad faster).
For large numbers (integers > 2^31), we recommend using bit64::integer64, even though the
default is to round off 0 bytes (full precision).
Value
setNumericRounding returns no value; the new value is applied. getNumericRounding returns the
current value: 0, 1 or 2.
See Also
datatable-optimize
http://en.wikipedia.org/wiki/Double-precision_floating-point_format
http://en.wikipedia.org/wiki/Floating_point
http://docs.oracle.com/cd/E19957-01/806-3568/ncg_goldberg.html
setops 87
Examples
DT = data.table(a=seq(0,1,by=0.2),b=1:2, key="a")
DT
setNumericRounding(0) # By default, rounding is turned off
DT[.(0.4)] # works
DT[.(0.6)] # no match, can be confusing since 0.6 is clearly there in DT
# happens due to floating point representation limitations
Description
Similar to base’s set functions, union, intersect, setdiff and setequal but for data.tables.
Additional all argument controls if/how duplicate rows are returned. bit64::integer64 is also
supported.
Unlike SQL, data.table functions will retain order of rows in result.
Usage
fintersect(x, y, all = FALSE)
fsetdiff(x, y, all = FALSE)
funion(x, y, all = FALSE)
fsetequal(x, y)
Arguments
x,y data.tables.
all Logical. Default is FALSE and removes duplicate rows on the result. When TRUE,
if there are xn copies of a particular row in x and yn copies of the same row in
y, then:
• fintersect will return min(xn, yn) copies of that row.
• fsetdiff will return max(0, xn-yn) copies of that row.
• funion will return xn+yn copies of that row.
88 setorder
Details
Columns of type complex and list are not supported except for funion.
Value
A data.table in case of fintersect, funion and fsetdiff. Logical TRUE or FALSE for fsetequal.
References
https://db.apache.org/derby/papers/Intersect-design.html
See Also
data.table, rbindlist, all.equal.data.table, unique, duplicated, uniqueN, anyDuplicated
Examples
x = data.table(c(1,2,2,2,3,4,4))
y = data.table(c(2,3,4,4,4,5))
fintersect(x, y) # intersect
fintersect(x, y, all=TRUE) # intersect all
fsetdiff(x, y) # except
fsetdiff(x, y, all=TRUE) # except all
funion(x, y) # union
funion(x, y, all=TRUE) # union all
fsetequal(x, y) # setequal
Description
In data.table parlance, all set* functions change their input by reference. That is, no copy is
made at all, other than temporary working memory, which is as large as one column. The only
other data.table operator that modifies input by reference is :=. Check out the See Also section
below for other set* function data.table provides.
setorder (and setorderv) reorders the rows of a data.table based on the columns (and column
order) provided. It reorders the table by reference and is therefore very memory efficient.
Also x[order(.)] is now optimised internally to use data.table’s fast order. data.table always
reorders in "C-locale" (see Details). To sort by session locale, use x[base::order(.)].
bit64::integer64 type is also supported for reordering rows of a data.table.
Usage
setorder(x, ..., na.last=FALSE)
setorderv(x, cols, order=1L, na.last=FALSE)
# optimised to use data.table's internal fast order
# x[order(., na.last=TRUE)]
setorder 89
Arguments
x A data.table.
... The columns to sort by. Do not quote column names. If ... is missing (ex:
setorder(x)), x is rearranged based on all columns in ascending order by de-
fault. To sort by a column in descending order prefix a "-", i.e., setorder(x, a, -b, c).
The -b works when b is of type character as well.
cols A character vector of column names of x, to which to order by. Do not add "-"
here. Use order argument instead.
order An integer vector with only possible values of 1 and -1, corresponding to as-
cending and descending order. The length of order must be either 1 or equal to
that of cols. If length(order) == 1, it’s recycled to length(cols).
na.last logical. If TRUE, missing values in the data are placed last; if FALSE, they are
placed first; if NA they are removed. na.last=NA is valid only for x[order(., na.last)]
and it’s default is TRUE. setorder and setorderv only accept TRUE/FALSE
with default FALSE.
Details
data.table implements fast radix based ordering. In versions <= 1.9.2, it was only capable of
increasing order (ascending). From 1.9.4 on, the functionality has been extended to decreasing
order (descending) as well.
setorder accepts unquoted column names (with names preceded with a - sign for descending
order) and reorders data.table rows by reference, for e.g., setorder(x, a, -b, c). Note that
-b also works with columns of type character unlike base::order, which requires -xtfrm(y)
instead (which is slow). setorderv in turn accepts a character vector of column names and an
integer vector of column order separately.
Note that setkey still requires and will always sort only in ascending order, and is different from
setorder in that it additionally sets the sorted attribute.
na.last argument, by default, is FALSE for setorder and setorderv to be consistent with data.table’s
setkey and is TRUE for x[order(.)] to be consistent with base::order. Only x[order(.)] can
have na.last = NA as it’s a subset operation as opposed to setorder or setorderv which reorders
the data.table by reference.
data.table always reorders in "C-locale". As a consequence, the ordering may be different
to that obtained by base::order. In English locales, for example, sorting is case-sensitive in
C-locale. Thus, sorting c("c", "a", "B") returns c("B", "a", "c") in data.table but
c("a", "B", "c") in base::order. Note this makes no difference in most cases of data; both re-
turn identical results on ids where only upper-case or lower-case letters are present ("AB123" < "AC234"
is true in both), or on country names and other proper nouns which are consistently capitalized. For
example, neither "America" < "Brazil" nor "america" < "brazil" are affected since the first
letter is consistently capitalized.
Using C-locale makes the behaviour of sorting in data.table more consistent across sessions
and locales. The behaviour of base::order depends on assumptions about the locale of the R
session. In English locales, "america" < "BRAZIL" is true by default but false if you either type
Sys.setlocale(locale="C") or the R session has been started in a C locale for you – which can
happen on servers/services since the locale comes from the environment the R session was started
90 shift
in. By contrast, "america" < "BRAZIL" is always false in data.table regardless of the way your
R session was started.
If setorder results in reordering of the rows of a keyed data.table, then it’s key will be set to
NULL.
Value
The input is modified by reference, and returned (invisibly) so it can be used in compound state-
ments; e.g., setorder(DT,a,-b)[, cumsum(c), by=list(a,b)]. If you require a copy, take a
copy first (using DT2 = copy(DT)). See ?copy.
See Also
setkey, setcolorder, setattr, setnames, set, :=, setDT, setDF, copy, setNumericRounding
Examples
set.seed(45L)
DT = data.table(A=sample(3, 10, TRUE),
B=sample(letters[1:3], 10, TRUE), C=sample(10))
# setorder
setorder(DT, A, -B)
Description
lead or lag vectors, lists, data.frames or data.tables implemented in C for speed.
bit64::integer64 is also supported.
Usage
shift(x, n=1L, fill=NA, type=c("lag", "lead"), give.names=FALSE)
Arguments
x A vector, list, data.frame or data.table.
n Non-negative integer vector denoting the offset to lead or lag the input by. To
create multiple lead/lag vectors, provide multiple values to n.
fill Value to pad by.
type default is "lag". The other possible value is "lead".
give.names default is FALSE which returns an unnamed list. When TRUE, names are auto-
matically generated corresponding to type and n.
shift 91
Details
shift accepts vectors, lists, data.frames or data.tables. It always returns a list except when the in-
put is a vector and length(n) == 1 in which case a vector is returned, for convenience. This is so
that it can be used conveniently within data.table’s syntax. For example, DT[, (cols) := shift(.SD, 1L), by=id]
would lag every column of .SD by 1 for each group and DT[, newcol := colA + shift(colB)]
would assign the sum of two vectors to newcol.
Argument n allows multiple values. For example, DT[, (cols) := shift(.SD, 1:2), by=id]
would lag every column of .SD by 1 and 2 for each group. If .SD contained four columns, the first
two elements of the list would correspond to lag=1 and lag=2 for the first column of .SD, the next
two for second column of .SD and so on. Please see examples for more.
shift is designed mainly for use in data.tables along with := or set. Therefore, it returns an un-
named list by default as assigning names for each group over and over can be quite time consuming
with many groups. It may be useful to set names automatically in other cases, which can be done
by setting give.names to TRUE.
Value
A list containing the lead/lag of input x.
See Also
data.table
Examples
# on vectors, returns a vector as long as length(n) == 1, #1127
x = 1:5
# lag with n=1 and pad with NA (returns vector)
shift(x, n=1, fill=NA, type="lag")
# lag with n=1 and 2, and pad with 0 (returns list)
shift(x, n=1:2, fill=0, type="lag")
# on data.tables
DT = data.table(year=2010:2014, v1=runif(5), v2=1:5, v3=letters[1:5])
# lag columns 'v1,v2,v3' DT by 1 and fill with 0
cols = c("v1","v2","v3")
anscols = paste("lead", cols, sep="_")
DT[, (anscols) := shift(.SD, 1, 0, "lead"), .SDcols=cols]
# while grouping
DT = data.table(year=rep(2010:2011, each=3), v1=1:6)
DT[, c("lag1", "lag2") := shift(.SD, 1:2), by=year]
# on lists
ll = list(1:3, letters[4:1], runif(2))
shift(ll, 1, type="lead")
shift(ll, 1, type="lead", give.names=TRUE)
shift(ll, 1:2, type="lead")
shouldPrint For use by packages that mimic/divert auto printing e.g. IRkernel and
knitr
Description
Not for use by users. Exported only for use by IRkernel (Jupyter) and knitr.
Usage
shouldPrint(x)
Arguments
x A data.table.
Details
Should IRkernel/Jupyter print a data.table returned invisibly by DT[,:=] ? This is a read-once func-
tion since it resets an internal flag. If you need the value more than once in your logic, store the
value from the first call.
Value
TRUE or FALSE.
References
https://github.com/IRkernel/IRkernel/issues/127
https://github.com/Rdatatable/data.table/issues/933
special-symbols 93
Description
.SD, .BY, .N, .I and .GRP are read only symbols for use in j. .N can be used in i as well. See the
vignettes and examples here and in data.table.
Details
The bindings of these variables are locked and attempting to assign to them will generate an error.
If you wish to manipulate .SD before returning it, take a copy(.SD) first (see FAQ 4.5). Using :=
in the j of .SD is reserved for future use as a (tortuously) flexible way to update DT by reference by
group (even when groups are not contiguous in an ad hoc by).
These symbols are used in j and defined as follows.
• .SD is a data.table containing the Subset of x’s Data for each group, excluding any columns
used in by (or keyby).
• .BY is a list containing a length 1 vector for each item in by. This can be useful when by is
not known in advance. The by variables are also available to j directly by name; useful for
example for titles of graphs if j is a plot command, or to branch with if() depending on the
value of a group variable.
• .N is an integer, length 1, containing the number of rows in the group. This may be useful
when the column names are not known in advance and for convenience generally. When
grouping by i, .N is the number of rows in x matched to, for each row of i, regardless of
whether nomatch is NA or 0. It is renamed to N (no dot) in the result (otherwise a column
called ".N" could conflict with the .N variable, see FAQ 4.6 for more details and example),
unless it is explicitly named; e.g., DT[,list(total=.N),by=a].
• .I is an integer vector equal to seq_len(nrow(x)). While grouping, it holds for each item in
the group, it’s row location in x. This is useful to subset in j; e.g. DT[, .I[which.max(somecol)], by=grp].
• .GRP is an integer, length 1, containing a simple group counter. 1 for the 1st group, 2 for the
2nd, etc.
See Also
data.table, :=, set, datatable-optimize
Examples
## Not run:
DT = data.table(x=rep(c("b","a","c"),each=3), v=c(1,1,1,2,2,1,1,2,2), y=c(1,3,6), a=1:9, b=9:1)
DT
X = data.table(x=c("c","b"), v=8:7, foo=c(4,2))
X
## End(Not run)
Description
Split method for data.table. Faster and more flexible. Be aware that processing list of data.tables
will be generally much slower than manipulation in single data.table by group using by argument,
read more on data.table.
Usage
## S3 method for class 'data.table'
split(x, f, drop = FALSE,
by, sorted = FALSE, keep.by = TRUE, flatten = TRUE,
..., verbose = getOption("datatable.verbose"))
Arguments
x data.table
f factor or list of factors. Same as split.data.frame. Use by argument instead,
this is just for consistency with data.frame method.
drop logical. Default FALSE will not drop empty list elements caused by factor levels
not referred by that factors. Works also with new arguments of split data.table
method.
by character vector. Column names on which split should be made. For length(by) > 1L
and flatten FALSE it will result nested lists with data.tables on leafs.
sorted When default FALSE it will retain the order of groups we are splitting on. When
TRUE then sorted list(s) are returned. Does not have effect for f argument.
keep.by logical default TRUE. Keep column provided to by argument.
flatten logical default TRUE will unlist nested lists of data.tables. When using f results
are always flattened to list of data.tables.
split 95
Details
Argument f is just for consistency in usage to data.frame method. Recommended is to use by
argument instead, it will be faster, more flexible, and by default will preserve order according to
order in data.
Value
List of data.tables. If using flatten FALSE and length(by) > 1L then recursively nested lists
having data.tables as leafs of grouping according to by argument.
See Also
data.table, rbindlist
Examples
set.seed(123)
DT = data.table(x1 = rep(letters[1:2], 6),
x2 = rep(letters[3:5], 4),
x3 = rep(letters[5:8], 3),
y = rnorm(12))
DT = DT[sample(.N)]
DF = as.data.frame(DT)
y = y)]
fdf = as.data.frame(fdt)
sdf = split(fdf, list(fdf$x1, fdf$x2))
all.equal(
split(fdt, by=c("x1", "x2"), sorted=TRUE),
lapply(sdf[sort(names(sdf))], setDT)
)
sdf = split(fdf, list(fdf$x1, fdf$x2), drop=TRUE)
all.equal(
split(fdt, by=c("x1", "x2"), sorted=TRUE, drop=TRUE),
lapply(sdf[sort(names(sdf))], setDT)
)
Description
Returns subsets of a data.table.
Usage
## S3 method for class 'data.table'
subset(x, subset, select, ...)
Arguments
x data.table to subset.
subset logical expression indicating elements or rows to keep
select expression indicating columns to select from data.table
... further arguments to be passed to or from other methods
Details
The subset argument works on the rows and will be evaluated in the data.table so columns can
be referred to (by name) as variables in the expression.
The data.table that is returned will maintain the original keys as long as they are not select-ed
out.
Value
A data.table containing the subset of rows and columns that are selected.
See Also
subset
tables 97
Examples
Description
Convenience function for concisely summarizing some metadata of all data.tables in memory (or
an optionally specified environment).
Usage
tables(mb=TRUE, order.col="NAME", width=80,
env=parent.frame(), silent=FALSE, index=FALSE)
Arguments
mb logical; TRUE adds the rough size of each data.table in megabytes to the
output under column MB.
order.col Column name (character) by which to sort the output.
width integer; number of characters beyond which the output for each of the columns
COLS, KEY, and INDICES are truncated.
env An environment, typically the .GlobalEnv by default, see Details.
silent logical; should the output be printed?
index logical; if TRUE, the column INDICES is added to indicate the indices assorted
with each object, see indices.
Details
Usually tables() is executed at the prompt, where parent.frame() returns .GlobalEnv. tables()
may also be useful inside functions where parent.frame() is the local scope of the function; in
such a scenario, simply set it to .GlobalEnv to get the same behaviour as at prompt.
Note that on older versions of R, object.size may be slow, so setting mb=FALSE may speed up
execution of tables significantly.
Setting silent=TRUE prints nothing; the metadata are returned as a data.table, invisibly, whether
silent is TRUE or FALSE.
98 test.data.table
Value
A data.table containing the information printed.
See Also
data.table, setkey, ls, objects, object.size
Examples
DT = data.table(A=1:10, B=letters[1:10])
DT2 = data.table(A=1:10000, ColB=10000:1)
setkey(DT,B)
tables()
Description
Runs a set of tests to check data.table is working correctly.
Usage
test.data.table(verbose=FALSE, pkg="pkg", silent=FALSE,
with.other.packages=FALSE, benchmark=FALSE)
Arguments
verbose If TRUE sets datatable.verbose to TRUE for the duration of the tests.
pkg Root directory name under which all package content (ex: DESCRIPTION, src/,
R/, inst/ etc..) resides.
silent Logical, default FALSE, when TRUE it will not raise error on in case of test fails.
with.other.packages
Run compatibility tests with other packages.
benchmark Run the benchmark script.
Details
Runs a series of tests. These can be used to see features and examples of usage, too. Running
test.data.table will tell you the full location of the test file(s) to open.
Value
When silent equals to TRUE it will return TRUE if all tests were successful. FALSE otherwise. If
silent equals to FALSE it will return TRUE if all tests were successful. Error otherwise.
timetaken 99
See Also
data.table
Examples
## Not run:
library(data.table)
test.data.table()
## End(Not run)
Description
Usage
timetaken(started.at)
Arguments
Value
Examples
started.at=proc.time()
Sys.sleep(1)
cat("Finished in",timetaken(started.at),"\n")
100 transform.data.table
Description
Usage
Arguments
Details
within is like with, but modifications (columns changed, added, or removed) are updated in the
returned data.table.
Note that transform will keep the key of the data.table provided the targets of the transform
(i.e. the columns that appear in . . . ) are not in the key of the data.table. within also retains the key
provided the key columns are not touched.
Value
See Also
Examples
DT <- data.table(a=rep(1:3, each=2), b=1:6)
DT$d = ave(DT$b, DT$c, FUN=max) # copies entire DT, even if it is 10GB in RAM
DT = DT[, transform(.SD, d=max(b)), by="c"] # same, but even worse as .SD is copied for each group
DT[, d:=max(b), by="c"] # same result, but much faster, shorter and scales
Description
Usage
Arguments
Details
The list elements (or columns of data.frame/data.table) should be all atomic. If list elements
are of unequal lengths, the value provided in fill will be used so that the resulting list always has
all elements of identical lengths. The class of input object is also preserved in the transposed result.
The ignore.empty argument can be used to skip or include length-0 elements.
This is particularly useful in tasks that require splitting a character column and assigning each part to
a separate column. This operation is quite common enough that a function tstrsplit is exported.
factor columns are converted to character type. Attributes are not preserved at the moment. This
may change in the future.
Value
A transposed list, data.frame or data.table.
See Also
data.table, tstrsplit
Examples
ll = list(1:5, 6:8)
transpose(ll)
setDT(transpose(ll, fill=0))[]
DT = data.table(x=1:5, y=6:10)
transpose(DT)
Description
These functions are experimental and somewhat advanced. By experimental we mean their names
might change and perhaps the syntax, argument names and types. So if you write a lot of code using
them, you have been warned! They should work and be stable, though, so please report problems
with them.
Usage
truelength(x)
alloc.col(DT,
n = getOption("datatable.alloccol"), # default: 1024L
verbose = getOption("datatable.verbose")) # default: FALSE
truelength 103
Arguments
Details
When adding columns by reference using :=, we could simply create a new column list vector
(one longer) and memcpy over the old vector, with no copy of the column vectors themselves.
That requires negligible use of space and time, and is what v1.7.2 did. However, that copy of the
list vector of column pointers only (but not the columns themselves), a shallow copy, resulted in
inconsistent behaviour in some circumstances. So, as from v1.7.3 data.table over allocates the list
vector of column pointers so that columns can be added fully by reference, consistently.
When the allocated column pointer slots are used up, to add a new column data.table must re-
allocate that vector. If two or more variables are bound to the same data.table this shallow copy
may or may not be desirable, but we don’t think this will be a problem very often (more discus-
sion may be required on data.table issue tracker). Setting options(datatable.verbose=TRUE)
includes messages if and when a shallow copy is taken. To avoid shallow copies there are several
options: use copy to make a deep copy first, use alloc.col to reallocate in advance, or, change the
default allocation rule (perhaps in your .Rprofile); e.g., options(datatable.alloccol=10000L).
Please note : over allocation of the column pointer vector is not for efficiency per se. It’s so that :=
can add columns by reference without a shallow copy.
Value
truelength(x) returns the length of the vector allocated in memory. length(x) of those items are
in use. Currently, it’s just the list vector of column pointers that is over-allocated (i.e. truelength(DT)),
not the column vectors themselves, which would in future allow fast row insert(). For tables
loaded from disk however, truelength is 0 in R 2.14.0+ (and random in R <= 2.13.2), which is
perhaps unexpected. data.table detects this state and over-allocates the loaded data.table when
the next column addition occurs. All other operations on data.table (such as fast grouping and
joins) do not need truelength.
alloc.col reallocates DT by reference. This may be useful for efficiency if you know you are about
to going to add a lot of columns in a loop. It also returns the new DT, for convenience in compound
queries.
See Also
copy
104 tstrsplit
Examples
DT = data.table(a=1:3,b=4:6)
length(DT) # 2 column pointer slots used
truelength(DT) # 1026 column pointer slots allocated
alloc.col(DT,2048)
length(DT) # 2 used
truelength(DT) # 2050 allocated, 2048 free
DT[,c:=7L] # add new column by assigning to spare slot
truelength(DT)-length(DT) # 2047 slots spare
Description
This is equivalent to transpose(strsplit(...)). This is a convenient wrapper function to split a
column using strsplit and assign the transposed result to individual columns. See examples.
Usage
tstrsplit(x, ..., fill=NA, type.convert=FALSE, keep, names=FALSE)
Arguments
x The vector to split (and transpose).
... All the arguments to be passed to strsplit.
fill Default is NA. It is used to fill shorter list elements so as to return each element
of the transposed result of equal lengths.
type.convert TRUE calls type.convert with as.is=TRUE on the columns.
keep Specify indices corresponding to just those list elements to retain in the trans-
posed result. Default is to return all.
names TRUE auto names the list with V1, V2 etc. Default (FALSE) is to return an un-
named list.
Details
It internally calls strsplit first, and then transpose on the result.
names argument can be used to return an auto named list, although this argument does not have
any effect when used with :=, which requires names to be provided explicitly. It might be useful in
other scenarios.
Value
A transposed list after splitting by the pattern provided.
update.dev.pkg 105
See Also
data.table, transpose
Examples
x = c("abcde", "ghij", "klmnopq")
strsplit(x, "", fixed=TRUE)
tstrsplit(x, "", fixed=TRUE)
tstrsplit(x, "", fixed=TRUE, fill="<NA>")
# names argument
tstrsplit(x, "", fixed=TRUE, keep=c(1,3,5), names=LETTERS[1:3])
Description
It will download and install package from devel repository only when new commit is available
there, otherwise only PACKAGES file is transferred. Defaults are set to update data.table, other
packages can be used. Their repository has to include git commit information in PACKAGES file.
Usage
## S3 method for class 'dev.pkg'
update(object="data.table",
repo="https://Rdatatable.github.io/data.table", field="Revision", ...)
Arguments
object character scalar, package name.
repo character scalar, url of package devel repository.
field character scalar, metadata field to use in PACKAGES file and DESCRIPTION
file, default "Revision".
... passed to install.packages.
Details
In case if devel repository does not provide package binaries user has have development tools in-
stalled for package compilation to use this function.
106 update.dev.pkg
Value
NULL.
See Also
data.table
Examples
# data.table::update.dev.pkg()
Index
107
108 INDEX
groupingsets, 55 na.omit.data.table, 68
NROW, 37, 62
haskey (setkey), 83
head, 37, 62 object.size, 98
hour (IDateTime), 57 objects, 98
Ops.data.table (data.table-package), 3
IDate (IDateTime), 57 order, 84
IDate-class (IDateTime), 57 order (setorder), 88
IDateTime, 8, 9, 57
index, 71 path.expand, 42
indices, 97 patterns, 63, 64, 69
indices (setkey), 83 print.data.table, 70
inrange (between), 23 print.default, 71
install.packages, 105 print.ITime (IDateTime), 57
integer64, 18
intersect (setops), 87 quarter (IDateTime), 57
is.data.table (as.data.table), 19
is.na.data.table (data.table-package), 3 rank, 41
isoweek (IDateTime), 57 rank (frank), 40
ITime (IDateTime), 57 rbind (rbindlist), 72
ITime-class (IDateTime), 57 rbindlist, 9, 20, 56, 72, 88, 95
read.csv, 47
J, 9, 20, 60, 85 rep.IDate (IDateTime), 57
rep.ITime (IDateTime), 57
key, 21, 71
rle, 74
key (setkey), 83
rleid, 9, 74, 75
key2 (setkey), 83
rleidv (rleid), 74
key<- (setkey), 83
rollup (groupingsets), 55
lag (shift), 90 round.IDate (IDateTime), 57
last, 37, 61 rounding (datatable.optimize), 28
lead (shift), 90 rowid, 9, 32, 74, 75
like, 24, 62 rowidv (rowid), 75
ls, 98 rownames, 22
tables, 9, 85, 97
tail, 37, 62
test.data.table, 9, 61, 98
timetaken, 99
transform, 100
transform (transform.data.table), 100
transform.data.table, 100
transpose, 101, 104, 105
truelength, 9, 15, 20, 102
tstrsplit, 102, 104
type.convert, 104
union (setops), 87
unique, 35, 88
unique (duplicated), 34
unique.data.frame, 35
unique.data.table, 9
uniqueN, 9, 88
uniqueN (duplicated), 34